This is page 19 of 25. Use http://codebase.md/beehiveinnovations/gemini-mcp-server?lines=true&page={x} to view the full context.
# Directory Structure
```
├── .claude
│ ├── commands
│ │ └── fix-github-issue.md
│ └── settings.json
├── .coveragerc
├── .dockerignore
├── .env.example
├── .gitattributes
├── .github
│ ├── FUNDING.yml
│ ├── ISSUE_TEMPLATE
│ │ ├── bug_report.yml
│ │ ├── config.yml
│ │ ├── documentation.yml
│ │ ├── feature_request.yml
│ │ └── tool_addition.yml
│ ├── pull_request_template.md
│ └── workflows
│ ├── docker-pr.yml
│ ├── docker-release.yml
│ ├── semantic-pr.yml
│ ├── semantic-release.yml
│ └── test.yml
├── .gitignore
├── .pre-commit-config.yaml
├── AGENTS.md
├── CHANGELOG.md
├── claude_config_example.json
├── CLAUDE.md
├── clink
│ ├── __init__.py
│ ├── agents
│ │ ├── __init__.py
│ │ ├── base.py
│ │ ├── claude.py
│ │ ├── codex.py
│ │ └── gemini.py
│ ├── constants.py
│ ├── models.py
│ ├── parsers
│ │ ├── __init__.py
│ │ ├── base.py
│ │ ├── claude.py
│ │ ├── codex.py
│ │ └── gemini.py
│ └── registry.py
├── code_quality_checks.ps1
├── code_quality_checks.sh
├── communication_simulator_test.py
├── conf
│ ├── __init__.py
│ ├── azure_models.json
│ ├── cli_clients
│ │ ├── claude.json
│ │ ├── codex.json
│ │ └── gemini.json
│ ├── custom_models.json
│ ├── dial_models.json
│ ├── gemini_models.json
│ ├── openai_models.json
│ ├── openrouter_models.json
│ └── xai_models.json
├── config.py
├── docker
│ ├── README.md
│ └── scripts
│ ├── build.ps1
│ ├── build.sh
│ ├── deploy.ps1
│ ├── deploy.sh
│ └── healthcheck.py
├── docker-compose.yml
├── Dockerfile
├── docs
│ ├── adding_providers.md
│ ├── adding_tools.md
│ ├── advanced-usage.md
│ ├── ai_banter.md
│ ├── ai-collaboration.md
│ ├── azure_openai.md
│ ├── configuration.md
│ ├── context-revival.md
│ ├── contributions.md
│ ├── custom_models.md
│ ├── docker-deployment.md
│ ├── gemini-setup.md
│ ├── getting-started.md
│ ├── index.md
│ ├── locale-configuration.md
│ ├── logging.md
│ ├── model_ranking.md
│ ├── testing.md
│ ├── tools
│ │ ├── analyze.md
│ │ ├── apilookup.md
│ │ ├── challenge.md
│ │ ├── chat.md
│ │ ├── clink.md
│ │ ├── codereview.md
│ │ ├── consensus.md
│ │ ├── debug.md
│ │ ├── docgen.md
│ │ ├── listmodels.md
│ │ ├── planner.md
│ │ ├── precommit.md
│ │ ├── refactor.md
│ │ ├── secaudit.md
│ │ ├── testgen.md
│ │ ├── thinkdeep.md
│ │ ├── tracer.md
│ │ └── version.md
│ ├── troubleshooting.md
│ ├── vcr-testing.md
│ └── wsl-setup.md
├── examples
│ ├── claude_config_macos.json
│ └── claude_config_wsl.json
├── LICENSE
├── providers
│ ├── __init__.py
│ ├── azure_openai.py
│ ├── base.py
│ ├── custom.py
│ ├── dial.py
│ ├── gemini.py
│ ├── openai_compatible.py
│ ├── openai.py
│ ├── openrouter.py
│ ├── registries
│ │ ├── __init__.py
│ │ ├── azure.py
│ │ ├── base.py
│ │ ├── custom.py
│ │ ├── dial.py
│ │ ├── gemini.py
│ │ ├── openai.py
│ │ ├── openrouter.py
│ │ └── xai.py
│ ├── registry_provider_mixin.py
│ ├── registry.py
│ ├── shared
│ │ ├── __init__.py
│ │ ├── model_capabilities.py
│ │ ├── model_response.py
│ │ ├── provider_type.py
│ │ └── temperature.py
│ └── xai.py
├── pyproject.toml
├── pytest.ini
├── README.md
├── requirements-dev.txt
├── requirements.txt
├── run_integration_tests.ps1
├── run_integration_tests.sh
├── run-server.ps1
├── run-server.sh
├── scripts
│ └── sync_version.py
├── server.py
├── simulator_tests
│ ├── __init__.py
│ ├── base_test.py
│ ├── conversation_base_test.py
│ ├── log_utils.py
│ ├── test_analyze_validation.py
│ ├── test_basic_conversation.py
│ ├── test_chat_simple_validation.py
│ ├── test_codereview_validation.py
│ ├── test_consensus_conversation.py
│ ├── test_consensus_three_models.py
│ ├── test_consensus_workflow_accurate.py
│ ├── test_content_validation.py
│ ├── test_conversation_chain_validation.py
│ ├── test_cross_tool_comprehensive.py
│ ├── test_cross_tool_continuation.py
│ ├── test_debug_certain_confidence.py
│ ├── test_debug_validation.py
│ ├── test_line_number_validation.py
│ ├── test_logs_validation.py
│ ├── test_model_thinking_config.py
│ ├── test_o3_model_selection.py
│ ├── test_o3_pro_expensive.py
│ ├── test_ollama_custom_url.py
│ ├── test_openrouter_fallback.py
│ ├── test_openrouter_models.py
│ ├── test_per_tool_deduplication.py
│ ├── test_planner_continuation_history.py
│ ├── test_planner_validation_old.py
│ ├── test_planner_validation.py
│ ├── test_precommitworkflow_validation.py
│ ├── test_prompt_size_limit_bug.py
│ ├── test_refactor_validation.py
│ ├── test_secaudit_validation.py
│ ├── test_testgen_validation.py
│ ├── test_thinkdeep_validation.py
│ ├── test_token_allocation_validation.py
│ ├── test_vision_capability.py
│ └── test_xai_models.py
├── systemprompts
│ ├── __init__.py
│ ├── analyze_prompt.py
│ ├── chat_prompt.py
│ ├── clink
│ │ ├── codex_codereviewer.txt
│ │ ├── default_codereviewer.txt
│ │ ├── default_planner.txt
│ │ └── default.txt
│ ├── codereview_prompt.py
│ ├── consensus_prompt.py
│ ├── debug_prompt.py
│ ├── docgen_prompt.py
│ ├── generate_code_prompt.py
│ ├── planner_prompt.py
│ ├── precommit_prompt.py
│ ├── refactor_prompt.py
│ ├── secaudit_prompt.py
│ ├── testgen_prompt.py
│ ├── thinkdeep_prompt.py
│ └── tracer_prompt.py
├── tests
│ ├── __init__.py
│ ├── CASSETTE_MAINTENANCE.md
│ ├── conftest.py
│ ├── gemini_cassettes
│ │ ├── chat_codegen
│ │ │ └── gemini25_pro_calculator
│ │ │ └── mldev.json
│ │ ├── chat_cross
│ │ │ └── step1_gemini25_flash_number
│ │ │ └── mldev.json
│ │ └── consensus
│ │ └── step2_gemini25_flash_against
│ │ └── mldev.json
│ ├── http_transport_recorder.py
│ ├── mock_helpers.py
│ ├── openai_cassettes
│ │ ├── chat_cross_step2_gpt5_reminder.json
│ │ ├── chat_gpt5_continuation.json
│ │ ├── chat_gpt5_moon_distance.json
│ │ ├── consensus_step1_gpt5_for.json
│ │ └── o3_pro_basic_math.json
│ ├── pii_sanitizer.py
│ ├── sanitize_cassettes.py
│ ├── test_alias_target_restrictions.py
│ ├── test_auto_mode_comprehensive.py
│ ├── test_auto_mode_custom_provider_only.py
│ ├── test_auto_mode_model_listing.py
│ ├── test_auto_mode_provider_selection.py
│ ├── test_auto_mode.py
│ ├── test_auto_model_planner_fix.py
│ ├── test_azure_openai_provider.py
│ ├── test_buggy_behavior_prevention.py
│ ├── test_cassette_semantic_matching.py
│ ├── test_challenge.py
│ ├── test_chat_codegen_integration.py
│ ├── test_chat_cross_model_continuation.py
│ ├── test_chat_openai_integration.py
│ ├── test_chat_simple.py
│ ├── test_clink_claude_agent.py
│ ├── test_clink_claude_parser.py
│ ├── test_clink_codex_agent.py
│ ├── test_clink_gemini_agent.py
│ ├── test_clink_gemini_parser.py
│ ├── test_clink_integration.py
│ ├── test_clink_parsers.py
│ ├── test_clink_tool.py
│ ├── test_collaboration.py
│ ├── test_config.py
│ ├── test_consensus_integration.py
│ ├── test_consensus_schema.py
│ ├── test_consensus.py
│ ├── test_conversation_continuation_integration.py
│ ├── test_conversation_field_mapping.py
│ ├── test_conversation_file_features.py
│ ├── test_conversation_memory.py
│ ├── test_conversation_missing_files.py
│ ├── test_custom_openai_temperature_fix.py
│ ├── test_custom_provider.py
│ ├── test_debug.py
│ ├── test_deploy_scripts.py
│ ├── test_dial_provider.py
│ ├── test_directory_expansion_tracking.py
│ ├── test_disabled_tools.py
│ ├── test_docker_claude_desktop_integration.py
│ ├── test_docker_config_complete.py
│ ├── test_docker_healthcheck.py
│ ├── test_docker_implementation.py
│ ├── test_docker_mcp_validation.py
│ ├── test_docker_security.py
│ ├── test_docker_volume_persistence.py
│ ├── test_file_protection.py
│ ├── test_gemini_token_usage.py
│ ├── test_image_support_integration.py
│ ├── test_image_validation.py
│ ├── test_integration_utf8.py
│ ├── test_intelligent_fallback.py
│ ├── test_issue_245_simple.py
│ ├── test_large_prompt_handling.py
│ ├── test_line_numbers_integration.py
│ ├── test_listmodels_restrictions.py
│ ├── test_listmodels.py
│ ├── test_mcp_error_handling.py
│ ├── test_model_enumeration.py
│ ├── test_model_metadata_continuation.py
│ ├── test_model_resolution_bug.py
│ ├── test_model_restrictions.py
│ ├── test_o3_pro_output_text_fix.py
│ ├── test_o3_temperature_fix_simple.py
│ ├── test_openai_compatible_token_usage.py
│ ├── test_openai_provider.py
│ ├── test_openrouter_provider.py
│ ├── test_openrouter_registry.py
│ ├── test_parse_model_option.py
│ ├── test_per_tool_model_defaults.py
│ ├── test_pii_sanitizer.py
│ ├── test_pip_detection_fix.py
│ ├── test_planner.py
│ ├── test_precommit_workflow.py
│ ├── test_prompt_regression.py
│ ├── test_prompt_size_limit_bug_fix.py
│ ├── test_provider_retry_logic.py
│ ├── test_provider_routing_bugs.py
│ ├── test_provider_utf8.py
│ ├── test_providers.py
│ ├── test_rate_limit_patterns.py
│ ├── test_refactor.py
│ ├── test_secaudit.py
│ ├── test_server.py
│ ├── test_supported_models_aliases.py
│ ├── test_thinking_modes.py
│ ├── test_tools.py
│ ├── test_tracer.py
│ ├── test_utf8_localization.py
│ ├── test_utils.py
│ ├── test_uvx_resource_packaging.py
│ ├── test_uvx_support.py
│ ├── test_workflow_file_embedding.py
│ ├── test_workflow_metadata.py
│ ├── test_workflow_prompt_size_validation_simple.py
│ ├── test_workflow_utf8.py
│ ├── test_xai_provider.py
│ ├── transport_helpers.py
│ └── triangle.png
├── tools
│ ├── __init__.py
│ ├── analyze.py
│ ├── apilookup.py
│ ├── challenge.py
│ ├── chat.py
│ ├── clink.py
│ ├── codereview.py
│ ├── consensus.py
│ ├── debug.py
│ ├── docgen.py
│ ├── listmodels.py
│ ├── models.py
│ ├── planner.py
│ ├── precommit.py
│ ├── refactor.py
│ ├── secaudit.py
│ ├── shared
│ │ ├── __init__.py
│ │ ├── base_models.py
│ │ ├── base_tool.py
│ │ ├── exceptions.py
│ │ └── schema_builders.py
│ ├── simple
│ │ ├── __init__.py
│ │ └── base.py
│ ├── testgen.py
│ ├── thinkdeep.py
│ ├── tracer.py
│ ├── version.py
│ └── workflow
│ ├── __init__.py
│ ├── base.py
│ ├── schema_builders.py
│ └── workflow_mixin.py
├── utils
│ ├── __init__.py
│ ├── client_info.py
│ ├── conversation_memory.py
│ ├── env.py
│ ├── file_types.py
│ ├── file_utils.py
│ ├── image_utils.py
│ ├── model_context.py
│ ├── model_restrictions.py
│ ├── security_config.py
│ ├── storage_backend.py
│ └── token_utils.py
└── zen-mcp-server
```
# Files
--------------------------------------------------------------------------------
/tools/codereview.py:
--------------------------------------------------------------------------------
```python
1 | """
2 | CodeReview Workflow tool - Systematic code review with step-by-step analysis
3 |
4 | This tool provides a structured workflow for comprehensive code review and analysis.
5 | It guides the CLI agent through systematic investigation steps with forced pauses between each step
6 | to ensure thorough code examination, issue identification, and quality assessment before proceeding.
7 | The tool supports complex review scenarios including security analysis, performance evaluation,
8 | and architectural assessment.
9 |
10 | Key features:
11 | - Step-by-step code review workflow with progress tracking
12 | - Context-aware file embedding (references during investigation, full content for analysis)
13 | - Automatic issue tracking with severity classification
14 | - Expert analysis integration with external models
15 | - Support for focused reviews (security, performance, architecture)
16 | - Confidence-based workflow optimization
17 | """
18 |
19 | import logging
20 | from typing import TYPE_CHECKING, Any, Literal, Optional
21 |
22 | from pydantic import Field, model_validator
23 |
24 | if TYPE_CHECKING:
25 | from tools.models import ToolModelCategory
26 |
27 | from config import TEMPERATURE_ANALYTICAL
28 | from systemprompts import CODEREVIEW_PROMPT
29 | from tools.shared.base_models import WorkflowRequest
30 |
31 | from .workflow.base import WorkflowTool
32 |
33 | logger = logging.getLogger(__name__)
34 |
35 | # Tool-specific field descriptions for code review workflow
36 | CODEREVIEW_WORKFLOW_FIELD_DESCRIPTIONS = {
37 | "step": (
38 | "Review narrative. Step 1: outline the review strategy. Later steps: report findings. MUST cover quality, security, "
39 | "performance, and architecture. Reference code via `relevant_files`; avoid dumping large snippets."
40 | ),
41 | "step_number": "Current review step (starts at 1) – each step should build on the last.",
42 | "total_steps": (
43 | "Number of review steps planned. External validation: two steps (analysis + summary). Internal validation: one step. "
44 | "Use the same limits when continuing an existing review via continuation_id."
45 | ),
46 | "next_step_required": (
47 | "True when another review step follows. External validation: step 1 → True, step 2 → False. Internal validation: set False immediately. "
48 | "Apply the same rule on continuation flows."
49 | ),
50 | "findings": "Capture findings (positive and negative) across quality, security, performance, and architecture; update each step.",
51 | "files_checked": "Absolute paths of every file reviewed, including those ruled out.",
52 | "relevant_files": "Step 1: list all files/dirs under review. Must be absolute full non-abbreviated paths. Final step: narrow to files tied to key findings.",
53 | "relevant_context": "Functions or methods central to findings (e.g. 'Class.method' or 'function_name').",
54 | "issues_found": "Issues with severity (critical/high/medium/low) and descriptions.",
55 | "review_validation_type": "Set 'external' (default) for expert follow-up or 'internal' for local-only review.",
56 | "images": "Optional diagram or screenshot paths that clarify review context.",
57 | "review_type": "Review focus: full, security, performance, or quick.",
58 | "focus_on": "Optional note on areas to emphasise (e.g. 'threading', 'auth flow').",
59 | "standards": "Coding standards or style guides to enforce.",
60 | "severity_filter": "Lowest severity to include when reporting issues (critical/high/medium/low/all).",
61 | }
62 |
63 |
64 | class CodeReviewRequest(WorkflowRequest):
65 | """Request model for code review workflow investigation steps"""
66 |
67 | # Required fields for each investigation step
68 | step: str = Field(..., description=CODEREVIEW_WORKFLOW_FIELD_DESCRIPTIONS["step"])
69 | step_number: int = Field(..., description=CODEREVIEW_WORKFLOW_FIELD_DESCRIPTIONS["step_number"])
70 | total_steps: int = Field(..., description=CODEREVIEW_WORKFLOW_FIELD_DESCRIPTIONS["total_steps"])
71 | next_step_required: bool = Field(..., description=CODEREVIEW_WORKFLOW_FIELD_DESCRIPTIONS["next_step_required"])
72 |
73 | # Investigation tracking fields
74 | findings: str = Field(..., description=CODEREVIEW_WORKFLOW_FIELD_DESCRIPTIONS["findings"])
75 | files_checked: list[str] = Field(
76 | default_factory=list, description=CODEREVIEW_WORKFLOW_FIELD_DESCRIPTIONS["files_checked"]
77 | )
78 | relevant_files: list[str] = Field(
79 | default_factory=list, description=CODEREVIEW_WORKFLOW_FIELD_DESCRIPTIONS["relevant_files"]
80 | )
81 | relevant_context: list[str] = Field(
82 | default_factory=list, description=CODEREVIEW_WORKFLOW_FIELD_DESCRIPTIONS["relevant_context"]
83 | )
84 | issues_found: list[dict] = Field(
85 | default_factory=list, description=CODEREVIEW_WORKFLOW_FIELD_DESCRIPTIONS["issues_found"]
86 | )
87 | # Deprecated confidence field kept for backward compatibility only
88 | confidence: Optional[str] = Field("low", exclude=True)
89 | review_validation_type: Optional[Literal["external", "internal"]] = Field(
90 | "external", description=CODEREVIEW_WORKFLOW_FIELD_DESCRIPTIONS.get("review_validation_type", "")
91 | )
92 |
93 | # Optional images for visual context
94 | images: Optional[list[str]] = Field(default=None, description=CODEREVIEW_WORKFLOW_FIELD_DESCRIPTIONS["images"])
95 |
96 | # Code review-specific fields (only used in step 1 to initialize)
97 | review_type: Optional[Literal["full", "security", "performance", "quick"]] = Field(
98 | "full", description=CODEREVIEW_WORKFLOW_FIELD_DESCRIPTIONS["review_type"]
99 | )
100 | focus_on: Optional[str] = Field(None, description=CODEREVIEW_WORKFLOW_FIELD_DESCRIPTIONS["focus_on"])
101 | standards: Optional[str] = Field(None, description=CODEREVIEW_WORKFLOW_FIELD_DESCRIPTIONS["standards"])
102 | severity_filter: Optional[Literal["critical", "high", "medium", "low", "all"]] = Field(
103 | "all", description=CODEREVIEW_WORKFLOW_FIELD_DESCRIPTIONS["severity_filter"]
104 | )
105 |
106 | # Override inherited fields to exclude them from schema (except model which needs to be available)
107 | temperature: Optional[float] = Field(default=None, exclude=True)
108 | thinking_mode: Optional[str] = Field(default=None, exclude=True)
109 |
110 | @model_validator(mode="after")
111 | def validate_step_one_requirements(self):
112 | """Ensure step 1 has required relevant_files field."""
113 | if self.step_number == 1 and not self.relevant_files:
114 | raise ValueError("Step 1 requires 'relevant_files' field to specify code files or directories to review")
115 | return self
116 |
117 |
118 | class CodeReviewTool(WorkflowTool):
119 | """
120 | Code Review workflow tool for step-by-step code review and expert analysis.
121 |
122 | This tool implements a structured code review workflow that guides users through
123 | methodical investigation steps, ensuring thorough code examination, issue identification,
124 | and quality assessment before reaching conclusions. It supports complex review scenarios
125 | including security audits, performance analysis, architectural review, and maintainability assessment.
126 | """
127 |
128 | def __init__(self):
129 | super().__init__()
130 | self.initial_request = None
131 | self.review_config = {}
132 |
133 | def get_name(self) -> str:
134 | return "codereview"
135 |
136 | def get_description(self) -> str:
137 | return (
138 | "Performs systematic, step-by-step code review with expert validation. "
139 | "Use for comprehensive analysis covering quality, security, performance, and architecture. "
140 | "Guides through structured investigation to ensure thoroughness."
141 | )
142 |
143 | def get_system_prompt(self) -> str:
144 | return CODEREVIEW_PROMPT
145 |
146 | def get_default_temperature(self) -> float:
147 | return TEMPERATURE_ANALYTICAL
148 |
149 | def get_model_category(self) -> "ToolModelCategory":
150 | """Code review requires thorough analysis and reasoning"""
151 | from tools.models import ToolModelCategory
152 |
153 | return ToolModelCategory.EXTENDED_REASONING
154 |
155 | def get_workflow_request_model(self):
156 | """Return the code review workflow-specific request model."""
157 | return CodeReviewRequest
158 |
159 | def get_input_schema(self) -> dict[str, Any]:
160 | """Generate input schema using WorkflowSchemaBuilder with code review-specific overrides."""
161 | from .workflow.schema_builders import WorkflowSchemaBuilder
162 |
163 | # Code review workflow-specific field overrides
164 | codereview_field_overrides = {
165 | "step": {
166 | "type": "string",
167 | "description": CODEREVIEW_WORKFLOW_FIELD_DESCRIPTIONS["step"],
168 | },
169 | "step_number": {
170 | "type": "integer",
171 | "minimum": 1,
172 | "description": CODEREVIEW_WORKFLOW_FIELD_DESCRIPTIONS["step_number"],
173 | },
174 | "total_steps": {
175 | "type": "integer",
176 | "minimum": 1,
177 | "description": CODEREVIEW_WORKFLOW_FIELD_DESCRIPTIONS["total_steps"],
178 | },
179 | "next_step_required": {
180 | "type": "boolean",
181 | "description": CODEREVIEW_WORKFLOW_FIELD_DESCRIPTIONS["next_step_required"],
182 | },
183 | "findings": {
184 | "type": "string",
185 | "description": CODEREVIEW_WORKFLOW_FIELD_DESCRIPTIONS["findings"],
186 | },
187 | "files_checked": {
188 | "type": "array",
189 | "items": {"type": "string"},
190 | "description": CODEREVIEW_WORKFLOW_FIELD_DESCRIPTIONS["files_checked"],
191 | },
192 | "relevant_files": {
193 | "type": "array",
194 | "items": {"type": "string"},
195 | "description": CODEREVIEW_WORKFLOW_FIELD_DESCRIPTIONS["relevant_files"],
196 | },
197 | "review_validation_type": {
198 | "type": "string",
199 | "enum": ["external", "internal"],
200 | "default": "external",
201 | "description": CODEREVIEW_WORKFLOW_FIELD_DESCRIPTIONS.get("review_validation_type", ""),
202 | },
203 | "issues_found": {
204 | "type": "array",
205 | "items": {"type": "object"},
206 | "description": CODEREVIEW_WORKFLOW_FIELD_DESCRIPTIONS["issues_found"],
207 | },
208 | "images": {
209 | "type": "array",
210 | "items": {"type": "string"},
211 | "description": CODEREVIEW_WORKFLOW_FIELD_DESCRIPTIONS["images"],
212 | },
213 | # Code review-specific fields (for step 1)
214 | "review_type": {
215 | "type": "string",
216 | "enum": ["full", "security", "performance", "quick"],
217 | "default": "full",
218 | "description": CODEREVIEW_WORKFLOW_FIELD_DESCRIPTIONS["review_type"],
219 | },
220 | "focus_on": {
221 | "type": "string",
222 | "description": CODEREVIEW_WORKFLOW_FIELD_DESCRIPTIONS["focus_on"],
223 | },
224 | "standards": {
225 | "type": "string",
226 | "description": CODEREVIEW_WORKFLOW_FIELD_DESCRIPTIONS["standards"],
227 | },
228 | "severity_filter": {
229 | "type": "string",
230 | "enum": ["critical", "high", "medium", "low", "all"],
231 | "default": "all",
232 | "description": CODEREVIEW_WORKFLOW_FIELD_DESCRIPTIONS["severity_filter"],
233 | },
234 | }
235 |
236 | # Use WorkflowSchemaBuilder with code review-specific tool fields
237 | return WorkflowSchemaBuilder.build_schema(
238 | tool_specific_fields=codereview_field_overrides,
239 | model_field_schema=self.get_model_field_schema(),
240 | auto_mode=self.is_effective_auto_mode(),
241 | tool_name=self.get_name(),
242 | )
243 |
244 | def get_required_actions(
245 | self, step_number: int, confidence: str, findings: str, total_steps: int, request=None
246 | ) -> list[str]:
247 | """Define required actions for each investigation phase.
248 |
249 | Now includes request parameter for continuation-aware decisions.
250 | """
251 | # Check for continuation - fast track mode
252 | if request:
253 | continuation_id = self.get_request_continuation_id(request)
254 | validation_type = self.get_review_validation_type(request)
255 | if continuation_id and validation_type == "external":
256 | if step_number == 1:
257 | return [
258 | "Quickly review the code files to understand context",
259 | "Identify any critical issues that need immediate attention",
260 | "Note main architectural patterns and design decisions",
261 | "Prepare summary of key findings for expert validation",
262 | ]
263 | else:
264 | return ["Complete review and proceed to expert analysis"]
265 |
266 | if step_number == 1:
267 | # Initial code review investigation tasks
268 | return [
269 | "Read and understand the code files specified for review",
270 | "Examine the overall structure, architecture, and design patterns used",
271 | "Identify the main components, classes, and functions in the codebase",
272 | "Understand the business logic and intended functionality",
273 | "Look for obvious issues: bugs, security concerns, performance problems",
274 | "Note any code smells, anti-patterns, or areas of concern",
275 | ]
276 | elif step_number == 2:
277 | # Deeper investigation for step 2
278 | return [
279 | "Examine specific code sections you've identified as concerning",
280 | "Analyze security implications: input validation, authentication, authorization",
281 | "Check for performance issues: algorithmic complexity, resource usage, inefficiencies",
282 | "Look for architectural problems: tight coupling, missing abstractions, scalability issues",
283 | "Identify code quality issues: readability, maintainability, error handling",
284 | "Search for over-engineering, unnecessary complexity, or design patterns that could be simplified",
285 | ]
286 | elif step_number >= 3:
287 | # Final verification for later steps
288 | return [
289 | "Verify all identified issues have been properly documented with severity levels",
290 | "Check for any missed critical security vulnerabilities or performance bottlenecks",
291 | "Confirm that architectural concerns and code quality issues are comprehensively captured",
292 | "Ensure positive aspects and well-implemented patterns are also noted",
293 | "Validate that your assessment aligns with the review type and focus areas specified",
294 | "Double-check that findings are actionable and provide clear guidance for improvements",
295 | ]
296 | else:
297 | # General investigation needed
298 | return [
299 | "Continue examining the codebase for additional patterns and potential issues",
300 | "Gather more evidence using appropriate code analysis techniques",
301 | "Test your assumptions about code behavior and design decisions",
302 | "Look for patterns that confirm or refute your current assessment",
303 | "Focus on areas that haven't been thoroughly examined yet",
304 | ]
305 |
306 | def should_call_expert_analysis(self, consolidated_findings, request=None) -> bool:
307 | """
308 | Decide when to call external model based on investigation completeness.
309 |
310 | For continuations with external type, always proceed with expert analysis.
311 | """
312 | # Check if user requested to skip assistant model
313 | if request and not self.get_request_use_assistant_model(request):
314 | return False
315 |
316 | # For continuations with external type, always proceed with expert analysis
317 | continuation_id = self.get_request_continuation_id(request)
318 | validation_type = self.get_review_validation_type(request)
319 | if continuation_id and validation_type == "external":
320 | return True # Always perform expert analysis for external continuations
321 |
322 | # Check if we have meaningful investigation data
323 | return (
324 | len(consolidated_findings.relevant_files) > 0
325 | or len(consolidated_findings.findings) >= 2
326 | or len(consolidated_findings.issues_found) > 0
327 | )
328 |
329 | def prepare_expert_analysis_context(self, consolidated_findings) -> str:
330 | """Prepare context for external model call for final code review validation."""
331 | context_parts = [
332 | f"=== CODE REVIEW REQUEST ===\\n{self.initial_request or 'Code review workflow initiated'}\\n=== END REQUEST ==="
333 | ]
334 |
335 | # Add investigation summary
336 | investigation_summary = self._build_code_review_summary(consolidated_findings)
337 | context_parts.append(
338 | f"\\n=== AGENT'S CODE REVIEW INVESTIGATION ===\\n{investigation_summary}\\n=== END INVESTIGATION ==="
339 | )
340 |
341 | # Add review configuration context if available
342 | if self.review_config:
343 | config_text = "\\n".join(f"- {key}: {value}" for key, value in self.review_config.items() if value)
344 | context_parts.append(f"\\n=== REVIEW CONFIGURATION ===\\n{config_text}\\n=== END CONFIGURATION ===")
345 |
346 | # Add relevant code elements if available
347 | if consolidated_findings.relevant_context:
348 | methods_text = "\\n".join(f"- {method}" for method in consolidated_findings.relevant_context)
349 | context_parts.append(f"\\n=== RELEVANT CODE ELEMENTS ===\\n{methods_text}\\n=== END CODE ELEMENTS ===")
350 |
351 | # Add issues found if available
352 | if consolidated_findings.issues_found:
353 | issues_text = "\\n".join(
354 | f"[{issue.get('severity', 'unknown').upper()}] {issue.get('description', 'No description')}"
355 | for issue in consolidated_findings.issues_found
356 | )
357 | context_parts.append(f"\\n=== ISSUES IDENTIFIED ===\\n{issues_text}\\n=== END ISSUES ===")
358 |
359 | # Add assessment evolution if available
360 | if consolidated_findings.hypotheses:
361 | assessments_text = "\\n".join(
362 | f"Step {h['step']} ({h['confidence']} confidence): {h['hypothesis']}"
363 | for h in consolidated_findings.hypotheses
364 | )
365 | context_parts.append(f"\\n=== ASSESSMENT EVOLUTION ===\\n{assessments_text}\\n=== END ASSESSMENTS ===")
366 |
367 | # Add images if available
368 | if consolidated_findings.images:
369 | images_text = "\\n".join(f"- {img}" for img in consolidated_findings.images)
370 | context_parts.append(
371 | f"\\n=== VISUAL REVIEW INFORMATION ===\\n{images_text}\\n=== END VISUAL INFORMATION ==="
372 | )
373 |
374 | return "\\n".join(context_parts)
375 |
376 | def _build_code_review_summary(self, consolidated_findings) -> str:
377 | """Prepare a comprehensive summary of the code review investigation."""
378 | summary_parts = [
379 | "=== SYSTEMATIC CODE REVIEW INVESTIGATION SUMMARY ===",
380 | f"Total steps: {len(consolidated_findings.findings)}",
381 | f"Files examined: {len(consolidated_findings.files_checked)}",
382 | f"Relevant files identified: {len(consolidated_findings.relevant_files)}",
383 | f"Code elements analyzed: {len(consolidated_findings.relevant_context)}",
384 | f"Issues identified: {len(consolidated_findings.issues_found)}",
385 | "",
386 | "=== INVESTIGATION PROGRESSION ===",
387 | ]
388 |
389 | for finding in consolidated_findings.findings:
390 | summary_parts.append(finding)
391 |
392 | return "\\n".join(summary_parts)
393 |
394 | def should_include_files_in_expert_prompt(self) -> bool:
395 | """Include files in expert analysis for comprehensive code review."""
396 | return True
397 |
398 | def should_embed_system_prompt(self) -> bool:
399 | """Embed system prompt in expert analysis for proper context."""
400 | return True
401 |
402 | def get_expert_thinking_mode(self) -> str:
403 | """Use high thinking mode for thorough code review analysis."""
404 | return "high"
405 |
406 | def get_expert_analysis_instruction(self) -> str:
407 | """Get specific instruction for code review expert analysis."""
408 | return (
409 | "Please provide comprehensive code review analysis based on the investigation findings. "
410 | "Focus on identifying any remaining issues, validating the completeness of the analysis, "
411 | "and providing final recommendations for code improvements, following the severity-based "
412 | "format specified in the system prompt."
413 | )
414 |
415 | # Hook method overrides for code review-specific behavior
416 |
417 | def prepare_step_data(self, request) -> dict:
418 | """
419 | Map code review-specific fields for internal processing.
420 | """
421 | step_data = {
422 | "step": request.step,
423 | "step_number": request.step_number,
424 | "findings": request.findings,
425 | "files_checked": request.files_checked,
426 | "relevant_files": request.relevant_files,
427 | "relevant_context": request.relevant_context,
428 | "issues_found": request.issues_found,
429 | "review_validation_type": self.get_review_validation_type(request),
430 | "hypothesis": request.findings, # Map findings to hypothesis for compatibility
431 | "images": request.images or [],
432 | "confidence": "high", # Dummy value for workflow_mixin compatibility
433 | }
434 | return step_data
435 |
436 | def should_skip_expert_analysis(self, request, consolidated_findings) -> bool:
437 | """
438 | Code review workflow skips expert analysis only when review_validation_type is "internal".
439 | Default is always to use expert analysis (external).
440 | For continuations with external type, always perform expert analysis immediately.
441 | """
442 | # If it's a continuation and review_validation_type is external, don't skip
443 | continuation_id = self.get_request_continuation_id(request)
444 | validation_type = self.get_review_validation_type(request)
445 | if continuation_id and validation_type != "internal":
446 | return False # Always do expert analysis for external continuations
447 |
448 | # Only skip if explicitly set to internal AND review is complete
449 | return validation_type == "internal" and not request.next_step_required
450 |
451 | def store_initial_issue(self, step_description: str):
452 | """Store initial request for expert analysis."""
453 | self.initial_request = step_description
454 |
455 | # Override inheritance hooks for code review-specific behavior
456 |
457 | def get_review_validation_type(self, request) -> str:
458 | """Get review validation type from request. Hook method for clean inheritance."""
459 | try:
460 | return request.review_validation_type or "external"
461 | except AttributeError:
462 | return "external" # Default to external validation
463 |
464 | def get_completion_status(self) -> str:
465 | """Code review tools use review-specific status."""
466 | return "code_review_complete_ready_for_implementation"
467 |
468 | def get_completion_data_key(self) -> str:
469 | """Code review uses 'complete_code_review' key."""
470 | return "complete_code_review"
471 |
472 | def get_final_analysis_from_request(self, request):
473 | """Code review tools use 'findings' field."""
474 | return request.findings
475 |
476 | def get_confidence_level(self, request) -> str:
477 | """Code review tools use 'certain' for high confidence."""
478 | return "certain"
479 |
480 | def get_completion_message(self) -> str:
481 | """Code review-specific completion message."""
482 | return (
483 | "Code review complete. You have identified all significant issues "
484 | "and provided comprehensive analysis. MANDATORY: Present the user with the complete review results "
485 | "categorized by severity, and IMMEDIATELY proceed with implementing the highest priority fixes "
486 | "or provide specific guidance for improvements. Focus on actionable recommendations."
487 | )
488 |
489 | def get_skip_reason(self) -> str:
490 | """Code review-specific skip reason."""
491 | return "Completed comprehensive code review with internal analysis only (no external model validation)"
492 |
493 | def get_skip_expert_analysis_status(self) -> str:
494 | """Code review-specific expert analysis skip status."""
495 | return "skipped_due_to_internal_analysis_type"
496 |
497 | def prepare_work_summary(self) -> str:
498 | """Code review-specific work summary."""
499 | return self._build_code_review_summary(self.consolidated_findings)
500 |
501 | def get_completion_next_steps_message(self, expert_analysis_used: bool = False) -> str:
502 | """
503 | Code review-specific completion message.
504 | """
505 | base_message = (
506 | "CODE REVIEW IS COMPLETE. You MUST now summarize and present ALL review findings organized by "
507 | "severity (Critical → High → Medium → Low), specific code locations with line numbers, and exact "
508 | "recommendations for improvement. Clearly prioritize the top 3 issues that need immediate attention. "
509 | "Provide concrete, actionable guidance for each issue—make it easy for a developer to understand "
510 | "exactly what needs to be fixed and how to implement the improvements."
511 | )
512 |
513 | # Add expert analysis guidance only when expert analysis was actually used
514 | if expert_analysis_used:
515 | expert_guidance = self.get_expert_analysis_guidance()
516 | if expert_guidance:
517 | return f"{base_message}\n\n{expert_guidance}"
518 |
519 | return base_message
520 |
521 | def get_expert_analysis_guidance(self) -> str:
522 | """
523 | Provide specific guidance for handling expert analysis in code reviews.
524 | """
525 | return (
526 | "IMPORTANT: Analysis from an assistant model has been provided above. You MUST critically evaluate and validate "
527 | "the expert findings rather than accepting them blindly. Cross-reference the expert analysis with "
528 | "your own investigation findings, verify that suggested improvements are appropriate for this "
529 | "codebase's context and patterns, and ensure recommendations align with the project's standards. "
530 | "Present a synthesis that combines your systematic review with validated expert insights, clearly "
531 | "distinguishing between findings you've independently confirmed and additional insights from expert analysis."
532 | )
533 |
534 | def get_step_guidance_message(self, request) -> str:
535 | """
536 | Code review-specific step guidance with detailed investigation instructions.
537 | """
538 | step_guidance = self.get_code_review_step_guidance(request.step_number, request)
539 | return step_guidance["next_steps"]
540 |
541 | def get_code_review_step_guidance(self, step_number: int, request) -> dict[str, Any]:
542 | """
543 | Provide step-specific guidance for code review workflow.
544 | Uses get_required_actions to determine what needs to be done,
545 | then formats those actions into appropriate guidance messages.
546 | """
547 | # Get the required actions from the single source of truth
548 | required_actions = self.get_required_actions(
549 | step_number,
550 | "medium", # Dummy value for backward compatibility
551 | request.findings or "",
552 | request.total_steps,
553 | request, # Pass request for continuation-aware decisions
554 | )
555 |
556 | # Check if this is a continuation to provide context-aware guidance
557 | continuation_id = self.get_request_continuation_id(request)
558 | validation_type = self.get_review_validation_type(request)
559 | is_external_continuation = continuation_id and validation_type == "external"
560 | is_internal_continuation = continuation_id and validation_type == "internal"
561 |
562 | # Step 1 handling
563 | if step_number == 1:
564 | if is_external_continuation:
565 | # Fast-track for external continuations
566 | return {
567 | "next_steps": (
568 | "You are on step 1 of MAXIMUM 2 steps for continuation. CRITICAL: Quickly review the code NOW. "
569 | "MANDATORY ACTIONS:\\n"
570 | + "\\n".join(f"{i+1}. {action}" for i, action in enumerate(required_actions))
571 | + "\\n\\nSet next_step_required=True and step_number=2 for the next call to trigger expert analysis."
572 | )
573 | }
574 | elif is_internal_continuation:
575 | # Internal validation mode
576 | next_steps = (
577 | "Continuing previous conversation with internal validation only. The analysis will build "
578 | "upon the prior findings without external model validation. REQUIRED ACTIONS:\\n"
579 | + "\\n".join(f"{i+1}. {action}" for i, action in enumerate(required_actions))
580 | )
581 | else:
582 | # Normal flow for new reviews
583 | next_steps = (
584 | f"MANDATORY: DO NOT call the {self.get_name()} tool again immediately. You MUST first examine "
585 | f"the code files thoroughly using appropriate tools. CRITICAL AWARENESS: You need to:\\n"
586 | + "\\n".join(f"{i+1}. {action}" for i, action in enumerate(required_actions))
587 | + f"\\n\\nOnly call {self.get_name()} again AFTER completing your investigation. "
588 | f"When you call {self.get_name()} next time, use step_number: {step_number + 1} "
589 | f"and report specific files examined, issues found, and code quality assessments discovered."
590 | )
591 |
592 | elif step_number == 2:
593 | # CRITICAL: Check if violating minimum step requirement
594 | if (
595 | request.total_steps >= 3
596 | and request.step_number < request.total_steps
597 | and not request.next_step_required
598 | ):
599 | next_steps = (
600 | f"ERROR: You set total_steps={request.total_steps} but next_step_required=False on step {request.step_number}. "
601 | f"This violates the minimum step requirement. You MUST set next_step_required=True until you reach the final step. "
602 | f"Call {self.get_name()} again with next_step_required=True and continue your investigation."
603 | )
604 | elif is_external_continuation or (not request.next_step_required and validation_type == "external"):
605 | # Fast-track completion or about to complete for external validation
606 | next_steps = (
607 | "Proceeding immediately to expert analysis. "
608 | f"MANDATORY: call {self.get_name()} tool immediately again, and set next_step_required=False to "
609 | f"trigger external validation NOW."
610 | )
611 | else:
612 | # Normal flow - deeper analysis needed
613 | next_steps = (
614 | f"STOP! Do NOT call {self.get_name()} again yet. You are on step 2 of {request.total_steps} minimum required steps. "
615 | f"MANDATORY ACTIONS before calling {self.get_name()} step {step_number + 1}:\\n"
616 | + "\\n".join(f"{i+1}. {action}" for i, action in enumerate(required_actions))
617 | + f"\\n\\nRemember: You MUST set next_step_required=True until step {request.total_steps}. "
618 | + f"Only call {self.get_name()} again with step_number: {step_number + 1} AFTER completing these code review tasks."
619 | )
620 |
621 | elif step_number >= 3:
622 | if not request.next_step_required and validation_type == "external":
623 | # About to complete - ready for expert analysis
624 | next_steps = (
625 | "Completing review and proceeding to expert analysis. "
626 | "Ensure all findings are documented with specific file references and line numbers."
627 | )
628 | else:
629 | # Later steps - final verification
630 | next_steps = (
631 | f"WAIT! Your code review needs final verification. DO NOT call {self.get_name()} immediately. REQUIRED ACTIONS:\\n"
632 | + "\\n".join(f"{i+1}. {action}" for i, action in enumerate(required_actions))
633 | + f"\\n\\nREMEMBER: Ensure you have identified all significant issues across all severity levels and "
634 | f"verified the completeness of your review. Document findings with specific file references and "
635 | f"line numbers where applicable, then call {self.get_name()} with step_number: {step_number + 1}."
636 | )
637 | else:
638 | # Fallback for any other case - check minimum step violation first
639 | if (
640 | request.total_steps >= 3
641 | and request.step_number < request.total_steps
642 | and not request.next_step_required
643 | ):
644 | next_steps = (
645 | f"ERROR: You set total_steps={request.total_steps} but next_step_required=False on step {request.step_number}. "
646 | f"This violates the minimum step requirement. You MUST set next_step_required=True until step {request.total_steps}."
647 | )
648 | elif not request.next_step_required and validation_type == "external":
649 | next_steps = (
650 | "Completing review. "
651 | "Ensure all findings are documented with specific file references and severity levels."
652 | )
653 | else:
654 | next_steps = (
655 | f"PAUSE REVIEW. Before calling {self.get_name()} step {step_number + 1}, you MUST examine more code thoroughly. "
656 | + "Required: "
657 | + ", ".join(required_actions[:2])
658 | + ". "
659 | + f"Your next {self.get_name()} call (step_number: {step_number + 1}) must include "
660 | f"NEW evidence from actual code analysis, not just theories. NO recursive {self.get_name()} calls "
661 | f"without investigation work!"
662 | )
663 |
664 | return {"next_steps": next_steps}
665 |
666 | def customize_workflow_response(self, response_data: dict, request) -> dict:
667 | """
668 | Customize response to match code review workflow format.
669 | """
670 | # Store initial request on first step
671 | if request.step_number == 1:
672 | self.initial_request = request.step
673 | # Store review configuration for expert analysis
674 | if request.relevant_files:
675 | self.review_config = {
676 | "relevant_files": request.relevant_files,
677 | "review_type": request.review_type,
678 | "focus_on": request.focus_on,
679 | "standards": request.standards,
680 | "severity_filter": request.severity_filter,
681 | }
682 |
683 | # Convert generic status names to code review-specific ones
684 | tool_name = self.get_name()
685 | status_mapping = {
686 | f"{tool_name}_in_progress": "code_review_in_progress",
687 | f"pause_for_{tool_name}": "pause_for_code_review",
688 | f"{tool_name}_required": "code_review_required",
689 | f"{tool_name}_complete": "code_review_complete",
690 | }
691 |
692 | if response_data["status"] in status_mapping:
693 | response_data["status"] = status_mapping[response_data["status"]]
694 |
695 | # Rename status field to match code review workflow
696 | if f"{tool_name}_status" in response_data:
697 | response_data["code_review_status"] = response_data.pop(f"{tool_name}_status")
698 | # Add code review-specific status fields
699 | response_data["code_review_status"]["issues_by_severity"] = {}
700 | for issue in self.consolidated_findings.issues_found:
701 | severity = issue.get("severity", "unknown")
702 | if severity not in response_data["code_review_status"]["issues_by_severity"]:
703 | response_data["code_review_status"]["issues_by_severity"][severity] = 0
704 | response_data["code_review_status"]["issues_by_severity"][severity] += 1
705 | response_data["code_review_status"]["review_validation_type"] = self.get_review_validation_type(request)
706 |
707 | # Map complete_codereviewworkflow to complete_code_review
708 | if f"complete_{tool_name}" in response_data:
709 | response_data["complete_code_review"] = response_data.pop(f"complete_{tool_name}")
710 |
711 | # Map the completion flag to match code review workflow
712 | if f"{tool_name}_complete" in response_data:
713 | response_data["code_review_complete"] = response_data.pop(f"{tool_name}_complete")
714 |
715 | return response_data
716 |
717 | # Required abstract methods from BaseTool
718 | def get_request_model(self):
719 | """Return the code review workflow-specific request model."""
720 | return CodeReviewRequest
721 |
722 | async def prepare_prompt(self, request) -> str:
723 | """Not used - workflow tools use execute_workflow()."""
724 | return "" # Workflow tools use execute_workflow() directly
725 |
```
--------------------------------------------------------------------------------
/CHANGELOG.md:
--------------------------------------------------------------------------------
```markdown
1 | # CHANGELOG
2 |
3 | <!-- version list -->
4 |
5 | ## v9.1.3 (2025-10-22)
6 |
7 | ### Bug Fixes
8 |
9 | - Reduced token usage, removed parameters from schema that CLIs never seem to use
10 | ([`3e27319`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/3e27319e60b0287df918856b58b2bbf042c948a8))
11 |
12 | - Telemetry option no longer available in gemini 0.11
13 | ([`2a8dff0`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/2a8dff0cc8a3f33111533cdb971d654637ed0578))
14 |
15 | ### Chores
16 |
17 | - Sync version to config.py [skip ci]
18 | ([`9e163f9`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/9e163f9dc0654fc28961c9897b7c787a2b96e57d))
19 |
20 | - Sync version to config.py [skip ci]
21 | ([`557e443`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/557e443a63ffd733fb41faaa8696f6f4bb2c2fd1))
22 |
23 | ### Refactoring
24 |
25 | - Improved precommit system prompt
26 | ([`3efff60`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/3efff6056e322ee1531d7bed5601038c129a8b29))
27 |
28 |
29 | ## v9.1.2 (2025-10-21)
30 |
31 | ### Bug Fixes
32 |
33 | - Configure codex with a longer timeout
34 | ([`d2773f4`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/d2773f488af28986632846652874de9ff633049c))
35 |
36 | - Handle claude's array style JSON https://github.com/BeehiveInnovations/zen-mcp-server/issues/295
37 | ([`d5790a9`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/d5790a9bfef719f03d17f2d719f1882e55d13b3b))
38 |
39 | ### Chores
40 |
41 | - Sync version to config.py [skip ci]
42 | ([`04132f1`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/04132f1459f1e086afd8e3d456f671b63338f846))
43 |
44 |
45 | ## v9.1.1 (2025-10-17)
46 |
47 | ### Bug Fixes
48 |
49 | - Failing test
50 | ([`aed3e3e`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/aed3e3ee80c440ac8ab0d4abbf235b84df723d18))
51 |
52 | - Handler for parsing multiple generated code blocks
53 | ([`f4c20d2`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/f4c20d2a20e1c57d8b10e8f508e07e2a8d72f94a))
54 |
55 | - Improved error reporting; codex cli would at times fail to figure out how to handle plain-text /
56 | JSON errors
57 | ([`95e69a7`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/95e69a7cb234305dcd37dcdd2f22be715922e9a8))
58 |
59 | ### Chores
60 |
61 | - Sync version to config.py [skip ci]
62 | ([`942757a`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/942757a360a74c021b2a1aa63e394f18f5abcecd))
63 |
64 |
65 | ## v9.1.0 (2025-10-17)
66 |
67 | ### Chores
68 |
69 | - Sync version to config.py [skip ci]
70 | ([`3ee0c8f`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/3ee0c8f555cb51b975700290919c2a8e2ada8cc4))
71 |
72 | ### Features
73 |
74 | - Enhance review prompts to emphasize static analysis
75 | ([`36e66e2`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/36e66e2e9a44a73a466545d4d3477ecb2cb3e669))
76 |
77 |
78 | ## v9.0.4 (2025-10-17)
79 |
80 | ### Chores
81 |
82 | - Sync version to config.py [skip ci]
83 | ([`8c6f653`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/8c6f6532d843f7f1b283ce9b6472e5ba991efe16))
84 |
85 |
86 | ## v9.0.3 (2025-10-16)
87 |
88 | ### Bug Fixes
89 |
90 | - Remove duplicate -o json flag in gemini CLI config
91 | ([`3b2eff5`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/3b2eff58ac0e2388045a7442c63f56ce259b54ba))
92 |
93 | ### Chores
94 |
95 | - Sync version to config.py [skip ci]
96 | ([`b205d71`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/b205d7159b674ce47ebc11af7255d1e3556fff93))
97 |
98 |
99 | ## v9.0.2 (2025-10-15)
100 |
101 | ### Bug Fixes
102 |
103 | - Update Claude CLI commands to new mcp syntax
104 | ([`a2189cb`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/a2189cb88a295ebad6268b9b08c893cd65bc1d89))
105 |
106 | ### Chores
107 |
108 | - Sync version to config.py [skip ci]
109 | ([`d08cdc6`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/d08cdc6691e0f68917f2824945905b7256e0e568))
110 |
111 |
112 | ## v9.0.1 (2025-10-14)
113 |
114 | ### Bug Fixes
115 |
116 | - Add JSON output flag to gemini CLI configuration
117 | ([`eb3dff8`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/eb3dff845828f60ff2659586883af622b8b035eb))
118 |
119 | ### Chores
120 |
121 | - Sync version to config.py [skip ci]
122 | ([`b9408aa`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/b9408aae8860d43b1da0ba67f9db98db7e4de2cf))
123 |
124 |
125 | ## v9.0.0 (2025-10-08)
126 |
127 | ### Chores
128 |
129 | - Sync version to config.py [skip ci]
130 | ([`23c9b35`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/23c9b35d5226b07b59a4c4b3d7833ba81b019ea8))
131 |
132 | ### Features
133 |
134 | - Claude Code as a CLI agent now supported. Mix and match: spawn claude code from within claude
135 | code, or claude code from within codex.
136 | ([`4cfaa0b`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/4cfaa0b6060769adfbd785a072526a5368421a73))
137 |
138 |
139 | ## v8.0.2 (2025-10-08)
140 |
141 | ### Bug Fixes
142 |
143 | - Restore run-server quote trimming regex
144 | ([`1de4542`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/1de454224c105891137134e2a25c2ee4f00dba45))
145 |
146 | ### Chores
147 |
148 | - Sync version to config.py [skip ci]
149 | ([`728fb43`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/728fb439b929c9dc37646b24537ae043208fda7d))
150 |
151 |
152 | ## v8.0.1 (2025-10-08)
153 |
154 | ### Bug Fixes
155 |
156 | - Resolve executable path for cross-platform compatibility in CLI agent
157 | ([`f98046c`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/f98046c2fccaa7f9a24665a0d705a98006461da5))
158 |
159 | ### Chores
160 |
161 | - Sync version to config.py [skip ci]
162 | ([`52245b9`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/52245b91eaa5d720f8c3b21ead55248dd8e8bd57))
163 |
164 | ### Testing
165 |
166 | - Fix clink agent tests to mock shutil.which() for executable resolution
167 | ([`4370be3`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/4370be33b4b69a40456527213bcd62321a925a57))
168 |
169 |
170 | ## v8.0.0 (2025-10-07)
171 |
172 | ### Chores
173 |
174 | - Sync version to config.py [skip ci]
175 | ([`4c34541`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/4c3454121c3c678cdfe8ea03fa77f4dd414df9bc))
176 |
177 |
178 | ## v7.8.1 (2025-10-07)
179 |
180 | ### Bug Fixes
181 |
182 | - Updated model description to fix test
183 | ([`04f7ce5`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/04f7ce5b03804564263f53a765931edba9c320cd))
184 |
185 | ### Chores
186 |
187 | - Sync version to config.py [skip ci]
188 | ([`c27e81d`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/c27e81d6d2f22978816f798a161a869d1ab5f025))
189 |
190 | ### Refactoring
191 |
192 | - Moved registries into a separate module and code cleanup
193 | ([`7c36b92`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/7c36b9255a13007a10af4fadefc21aadfce482b0))
194 |
195 |
196 | ## v7.8.0 (2025-10-07)
197 |
198 | ### Chores
199 |
200 | - Sync version to config.py [skip ci]
201 | ([`3e5fa96`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/3e5fa96c981bbd7b844a9887a518ffe266b78e9b))
202 |
203 | ### Documentation
204 |
205 | - Consensus video
206 | ([`2352684`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/23526841922a73c68094e5205e19af04a1f6c8cc))
207 |
208 | - Formatting
209 | ([`7d7c74b`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/7d7c74b5a38b7d1adf132b8e28034017df7aa852))
210 |
211 | - Link to videos from main page
212 | ([`e8ef193`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/e8ef193daba393b55a3beaaba49721bb9182378a))
213 |
214 | - Update README.md
215 | ([`7b13543`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/7b13543824fc0af729daf753ecdddba9ee7d9f1e))
216 |
217 | ### Features
218 |
219 | - All native providers now read from catalog files like OpenRouter / Custom configs. Allows for
220 | greater control over the capabilities
221 | ([`2a706d5`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/2a706d5720c0bf97b71c3e0fc95c15f78015bedf))
222 |
223 | - Provider cleanup
224 | ([`9268dda`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/9268ddad2a07306351765b47098134512739f49f))
225 |
226 | ### Refactoring
227 |
228 | - New base class for model registry / loading
229 | ([`02d13da`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/02d13da897016d7491b4a10a1195983385d66654))
230 |
231 |
232 | ## v7.7.0 (2025-10-07)
233 |
234 | ### Chores
235 |
236 | - Sync version to config.py [skip ci]
237 | ([`70ae62a`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/70ae62a2cd663c3abcabddd1be1bc6ed9512d7df))
238 |
239 | ### Documentation
240 |
241 | - Video
242 | ([`ed5dda7`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/ed5dda7c5a9439c2835cc69d76e6377169ad048a))
243 |
244 | ### Features
245 |
246 | - More aliases
247 | ([`5f0aaf5`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/5f0aaf5f69c9d188d817b5ffbf6738c61da40ec7))
248 |
249 |
250 | ## v7.6.0 (2025-10-07)
251 |
252 | ### Chores
253 |
254 | - Sync version to config.py [skip ci]
255 | ([`c1c75ba`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/c1c75ba304c2840329650c46273e87eab9b05906))
256 |
257 | - Sync version to config.py [skip ci]
258 | ([`0fa9b66`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/0fa9b6658099c8e0d79fda0c7d2347f62d0e6137))
259 |
260 | ### Documentation
261 |
262 | - Info about AI client timeouts
263 | ([`3ddfed5`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/3ddfed5ef09000791e1c94b041c43dc273ed53a8))
264 |
265 | ### Features
266 |
267 | - Add support for openai/gpt-5-pro model
268 | ([`abed075`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/abed075b2eaa99e9618202f47ff921094baae952))
269 |
270 |
271 | ## v7.5.2 (2025-10-06)
272 |
273 | ### Bug Fixes
274 |
275 | - Handle 429 response https://github.com/BeehiveInnovations/zen-mcp-server/issues/273
276 | ([`cbe1d79`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/cbe1d7993276bd014b495cbd2d0ece1f5d7583d9))
277 |
278 | ### Chores
279 |
280 | - Sync version to config.py [skip ci]
281 | ([`74fdd36`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/74fdd36de92d34681fcc5a2f772c3d05634f0a55))
282 |
283 |
284 | ## v7.5.1 (2025-10-06)
285 |
286 | ### Chores
287 |
288 | - Sync version to config.py [skip ci]
289 | ([`004e379`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/004e379cf2f1853829dccb15fa72ec18d282f1a4))
290 |
291 |
292 | ## v7.5.0 (2025-10-06)
293 |
294 | ### Chores
295 |
296 | - Sync version to config.py [skip ci]
297 | ([`71e7cd5`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/71e7cd55b1f4955a6d718fddc0de419414d133b6))
298 |
299 | ### Documentation
300 |
301 | - Video
302 | ([`775e4d5`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/775e4d50b826858095c5f2a61a07fc01c4a00816))
303 |
304 | - Videos
305 | ([`bb2066c`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/bb2066c909f6581ba40fc5ddef3870954ae553ab))
306 |
307 | ### Features
308 |
309 | - Support for GPT-5-Pro highest reasoning model
310 | https://github.com/BeehiveInnovations/zen-mcp-server/issues/275
311 | ([`a65485a`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/a65485a1e52fc79739000426295a27d096f4c9d8))
312 |
313 |
314 | ## v7.4.0 (2025-10-06)
315 |
316 | ### Chores
317 |
318 | - Sync version to config.py [skip ci]
319 | ([`76bf98e`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/76bf98e5cd972dabd3c79b25fcb9b9a717b23f6d))
320 |
321 | ### Features
322 |
323 | - Improved prompt
324 | ([`b1e9963`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/b1e9963991a41dff082ec1dce5691c318f105e6d))
325 |
326 |
327 | ## v7.3.0 (2025-10-06)
328 |
329 | ### Chores
330 |
331 | - Sync version to config.py [skip ci]
332 | ([`e7920d0`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/e7920d0ed16c0e6de9d1ccaa0b58d3fb5cbd7f2f))
333 |
334 | ### Documentation
335 |
336 | - Fixed typo
337 | ([`3ab0aa8`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/3ab0aa8314ad5992bcb00de549a0fab2e522751d))
338 |
339 | - Fixed typo
340 | ([`c17ce3c`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/c17ce3cf958d488b97fa7127942542ab514b58bd))
341 |
342 | - Update apilookup.md
343 | ([`1918679`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/19186794edac4fce5523e671310aecff4cbfdc81))
344 |
345 | - Update README.md
346 | ([`23c6c78`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/23c6c78bf152ede6e7b5f7b7770b12a8442845a3))
347 |
348 | ### Features
349 |
350 | - Codex supports web-search natively but needs to be turned on, run-server script asks if the user
351 | would like this done
352 | ([`97ba7e4`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/97ba7e44ce7e3fd874759514ed2f0738033fc801))
353 |
354 |
355 | ## v7.2.0 (2025-10-06)
356 |
357 | ### Chores
358 |
359 | - Sync version to config.py [skip ci]
360 | ([`1854b1e`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/1854b1e26b705cda0dc3f4d733647f1454aa0352))
361 |
362 | ### Documentation
363 |
364 | - Updated
365 | ([`bb57f71`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/bb57f719666ab6a586d835688ff8086282a5a0dc))
366 |
367 | ### Features
368 |
369 | - New tool to perform apilookup (latest APIs / SDKs / language features etc)
370 | https://github.com/BeehiveInnovations/zen-mcp-server/issues/204
371 | ([`5bea595`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/5bea59540f58b3c45044828c10f131aed104dd1c))
372 |
373 | ### Refactoring
374 |
375 | - De-duplicate roles to avoid explosion when more CLIs get added
376 | ([`c42e9e9`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/c42e9e9c34d7ae4732e2e4fbed579b681a6d170d))
377 |
378 |
379 | ## v7.1.1 (2025-10-06)
380 |
381 | ### Bug Fixes
382 |
383 | - Clink missing in toml
384 | ([`1ff77fa`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/1ff77faa800ad6c2dde49cad98dfa72035fe1c81))
385 |
386 | ### Chores
387 |
388 | - Sync version to config.py [skip ci]
389 | ([`e02e78d`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/e02e78d903b35f4c01b8039f4157e97b38d3ec7b))
390 |
391 | ### Documentation
392 |
393 | - Example for codex cli
394 | ([`344c42b`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/344c42bcbfb543bfd05cbc27fd5b419c76b77954))
395 |
396 | - Example for codex cli
397 | ([`c3044de`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/c3044de7424e638dde5c8ec49adb6c3c7c5a60b2))
398 |
399 | - Update README.md
400 | ([`2e719ae`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/2e719ae35e7979f7b83bd910867e79863a7f9ceb))
401 |
402 |
403 | ## v7.1.0 (2025-10-05)
404 |
405 | ### Chores
406 |
407 | - Sync version to config.py [skip ci]
408 | ([`d54bfdd`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/d54bfdd49797d076ec9cade44c56292a8089c744))
409 |
410 | ### Features
411 |
412 | - Support for codex as external CLI
413 | ([`561e4aa`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/561e4aaaa8a89eb89c03985b9e7720cc98ef666c))
414 |
415 |
416 | ## v7.0.2 (2025-10-05)
417 |
418 | ### Chores
419 |
420 | - Sync version to config.py [skip ci]
421 | ([`f2142a2`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/f2142a22ec50abc54b464eedd6b8239d20c509be))
422 |
423 |
424 | ## v7.0.1 (2025-10-05)
425 |
426 | ### Bug Fixes
427 |
428 | - --yolo needed for running shell commands, documentation added
429 | ([`15ae3f2`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/15ae3f24babccf42f43be5028bf8c60c05a6beaf))
430 |
431 | ### Chores
432 |
433 | - Sync version to config.py [skip ci]
434 | ([`bc4a27b`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/bc4a27b18a4a3f45afb22178e61ea0be4d6a273c))
435 |
436 | ### Documentation
437 |
438 | - Updated intro
439 | ([`fb668c3`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/fb668c39b5f6e3dd37f7027f953f6004f258f2bf))
440 |
441 |
442 | ## v7.0.0 (2025-10-05)
443 |
444 | ### Chores
445 |
446 | - Sync version to config.py [skip ci]
447 | ([`0d46976`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/0d46976a8aa85254e4dbe06f5e71161cd3b13938))
448 |
449 | - Sync version to config.py [skip ci]
450 | ([`8296bf8`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/8296bf871c39597a904c70e7d98c72fcb4dc5a84))
451 |
452 | ### Documentation
453 |
454 | - Instructions for OpenCode
455 | ([`bd66622`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/bd666227c8f7557483f7e24fb8544fc0456600dc))
456 |
457 | - Updated intro
458 | ([`615873c`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/615873c3db2ecf5ce6475caa3445e1da9a2517bd))
459 |
460 | ### Features
461 |
462 | - Huge update - Link another CLI (such as `gemini` directly from with Claude Code / Codex).
463 | https://github.com/BeehiveInnovations/zen-mcp-server/issues/208
464 | ([`a2ccb48`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/a2ccb48e9a5080a75dbfd483b5f09fc719c887e5))
465 |
466 | ### Refactoring
467 |
468 | - Fixed test
469 | ([`9c99b9b`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/9c99b9b35219f54db8d7be0958d4390a106631ae))
470 |
471 | - Include file modification dates too
472 | ([`47973e9`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/47973e945efa2cdbdb8f3404d467d7f1abc62b0a))
473 |
474 |
475 | ## v6.1.0 (2025-10-04)
476 |
477 | ### Chores
478 |
479 | - Sync version to config.py [skip ci]
480 | ([`18095d7`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/18095d7d398e4bf3d24c57a52c81ac619acb1b89))
481 |
482 | ### Documentation
483 |
484 | - Updated intro
485 | ([`aa65394`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/aa6539472c4ddf1c3c1bac446fdee03e75e1cb50))
486 |
487 | ### Features
488 |
489 | - Support for Qwen Code
490 | ([`fe9968b`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/fe9968b633d0312b82426e9ebddfe1d6515be3c5))
491 |
492 |
493 | ## v6.0.0 (2025-10-04)
494 |
495 | ### Chores
496 |
497 | - Sync version to config.py [skip ci]
498 | ([`ae8749a`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/ae8749ab37bdaa7e225b5219820adeb74ca9a552))
499 |
500 | ### Documentation
501 |
502 | - Updated
503 | ([`e91ed2a`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/e91ed2a924b1702edf9e1417479ac0dee0ca1553))
504 |
505 | ### Features
506 |
507 | - Azure OpenAI / Azure AI Foundry support. Models should be defined in conf/azure_models.json (or a
508 | custom path). See .env.example for environment variables or see readme.
509 | https://github.com/BeehiveInnovations/zen-mcp-server/issues/265
510 | ([`ff9a07a`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/ff9a07a37adf7a24aa87c63b3ba9db88bdff467b))
511 |
512 | - Breaking change - OpenRouter models are now read from conf/openrouter_models.json while Custom /
513 | Self-hosted models are read from conf/custom_models.json
514 | ([`ff9a07a`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/ff9a07a37adf7a24aa87c63b3ba9db88bdff467b))
515 |
516 | - OpenAI/compatible models (such as Azure OpenAI) can declare if they use the response API instead
517 | via `use_openai_responses_api`
518 | ([`3824d13`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/3824d131618683572e9e8fffa6b25ccfabf4cf50))
519 |
520 | - OpenRouter / Custom Models / Azure can separately also use custom config paths now (see
521 | .env.example )
522 | ([`ff9a07a`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/ff9a07a37adf7a24aa87c63b3ba9db88bdff467b))
523 |
524 | ### Refactoring
525 |
526 | - Breaking change: `is_custom` property has been removed from model_capabilities.py (and thus
527 | custom_models.json) given each models are now read from separate configuration files
528 | ([`ff9a07a`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/ff9a07a37adf7a24aa87c63b3ba9db88bdff467b))
529 |
530 | - Model registry class made abstract, OpenRouter / Custom Provider / Azure OpenAI now subclass these
531 | ([`ff9a07a`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/ff9a07a37adf7a24aa87c63b3ba9db88bdff467b))
532 |
533 |
534 | ## v5.22.0 (2025-10-04)
535 |
536 | ### Bug Fixes
537 |
538 | - CI test
539 | ([`bc93b53`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/bc93b5343bbd8657b95ab47c00a2cb99a68a009f))
540 |
541 | - Listmodels to always honor restricted models
542 | ([`4015e91`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/4015e917ed32ae374ec6493b74993fcb34f4a971))
543 |
544 | ### Chores
545 |
546 | - Sync version to config.py [skip ci]
547 | ([`054e34e`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/054e34e31ca5bee5a11c0e3e6537f58e8897c79c))
548 |
549 | - Sync version to config.py [skip ci]
550 | ([`c0334d7`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/c0334d77922f1b05e3fd755851da112567fb9ae6))
551 |
552 | ### Features
553 |
554 | - Centralized environment handling, ensures ZEN_MCP_FORCE_ENV_OVERRIDE is honored correctly
555 | ([`2c534ac`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/2c534ac06e4c6078b96781dfb55c5759b982afe8))
556 |
557 | ### Refactoring
558 |
559 | - Don't retry on 429
560 | ([`d184024`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/d18402482087f52b7bd07755c9304ed00ed20592))
561 |
562 | - Improved retry logic and moved core logic to base class
563 | ([`f955100`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/f955100f3a82973ccd987607e1d8a1bbe07828c8))
564 |
565 | - Removed subclass override when the base class should be resolving the model name
566 | ([`06d7701`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/06d7701cc3ee09732ab713fa9c7c004199154483))
567 |
568 |
569 | ## v5.21.0 (2025-10-03)
570 |
571 | ### Chores
572 |
573 | - Sync version to config.py [skip ci]
574 | ([`ddb20a6`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/ddb20a6cdb8cdeee27c0aacb0b9c794283b5774c))
575 |
576 |
577 | ## v5.20.1 (2025-10-03)
578 |
579 | ### Chores
580 |
581 | - Sync version to config.py [skip ci]
582 | ([`03addcf`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/03addcfa2d3aed5086fe4c94e8b9ae56229a93ae))
583 |
584 |
585 | ## v5.20.0 (2025-10-03)
586 |
587 | ### Chores
588 |
589 | - Sync version to config.py [skip ci]
590 | ([`539bc72`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/539bc72f1ca2a2cadcccad02de1fd5fc22cd0415))
591 |
592 |
593 | ## v5.19.0 (2025-10-03)
594 |
595 | ### Bug Fixes
596 |
597 | - Add GPT-5-Codex to Responses API routing and simplify comments
598 | ([`82b021d`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/82b021d75acc791e68c7afb35f6492f68cf02bec))
599 |
600 | ### Chores
601 |
602 | - Sync version to config.py [skip ci]
603 | ([`8e32ef3`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/8e32ef33e3ce7ab2a9d7eb5c90fe5b93b12d5c26))
604 |
605 | ### Documentation
606 |
607 | - Bumped defaults
608 | ([`95d98a9`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/95d98a9bc0a5bafadccb9f6d1e4eda97a0dd2ce7))
609 |
610 | ### Features
611 |
612 | - Add GPT-5-Codex support with Responses API integration
613 | ([`f265342`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/f2653427ca829368e7145325d20a98df3ee6d6b4))
614 |
615 | ### Testing
616 |
617 | - Cross tool memory recall, testing continuation via cassette recording
618 | ([`88493bd`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/88493bd357c6a12477c3160813100dae1bc46493))
619 |
620 |
621 | ## v5.18.3 (2025-10-03)
622 |
623 | ### Bug Fixes
624 |
625 | - External model name now recorded properly in responses
626 | ([`d55130a`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/d55130a430401e106cd86f3e830b3d756472b7ff))
627 |
628 | ### Chores
629 |
630 | - Sync version to config.py [skip ci]
631 | ([`5714e20`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/5714e2016405f7607b44d78f85081c7ccee706e5))
632 |
633 | ### Documentation
634 |
635 | - Updated docs
636 | ([`b4e5090`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/b4e50901ba60c88137a29d00ecf99718582856d3))
637 |
638 | ### Refactoring
639 |
640 | - Generic name for the CLI agent
641 | ([`e9b6947`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/e9b69476cd922c12931d62ccc3be9082bbbf6014))
642 |
643 | - Generic name for the CLI agent
644 | ([`7a6fa0e`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/7a6fa0e77a8c4a682dc11c9bbb16bdaf86d9edf4))
645 |
646 | - Generic name for the CLI agent
647 | ([`b692da2`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/b692da2a82facce7455b8f2ec0108e1db84c07c3))
648 |
649 | - Generic name for the CLI agent
650 | ([`f76ebbf`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/f76ebbf280cc78ffcfe17cb4590aeaa231db8aa1))
651 |
652 | - Generic name for the CLI agent
653 | ([`c05913a`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/c05913a09e53e195b9a108647c09c061ced19d17))
654 |
655 | - Generic name for the CLI agent
656 | ([`0dfaa63`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/0dfaa6312ed95ac3d1ae0032334ae1286871b15e))
657 |
658 | ### Testing
659 |
660 | - Fixed integration tests, removed magicmock
661 | ([`87ccb6b`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/87ccb6b25ba32a3cb9c4cc64fc0e96294f492c04))
662 |
663 |
664 | ## v5.18.2 (2025-10-02)
665 |
666 | ### Bug Fixes
667 |
668 | - Https://github.com/BeehiveInnovations/zen-mcp-server/issues/194
669 | ([`8b3a286`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/8b3a2867fb83eccb3a8e8467e7e3fc5b8ebe1d0c))
670 |
671 | ### Chores
672 |
673 | - Sync version to config.py [skip ci]
674 | ([`bf2196c`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/bf2196cdd58ae8d8d93597f2be69c798265d678f))
675 |
676 |
677 | ## v5.18.1 (2025-10-02)
678 |
679 | ### Chores
680 |
681 | - Sync version to config.py [skip ci]
682 | ([`e434a26`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/e434a2614af82efd15de4dd94b2c30559c91414e))
683 |
684 |
685 | ## v5.18.0 (2025-10-02)
686 |
687 | ### Chores
688 |
689 | - Sync version to config.py [skip ci]
690 | ([`e78fe35`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/e78fe35a1b64cc0ed89664440ef7c7b94495d7dc))
691 |
692 | ### Features
693 |
694 | - Added `intelligence_score` to the model capabilities schema; a 1-20 number that can be specified
695 | to influence the sort order of models presented to the CLI in `auto selection` mode
696 | ([`6cab9e5`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/6cab9e56fc5373da5c11d4545bcb85371d4803a4))
697 |
698 |
699 | ## v5.17.4 (2025-10-02)
700 |
701 | ### Chores
702 |
703 | - Sync version to config.py [skip ci]
704 | ([`a6c9b92`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/a6c9b9212c77852d9e9a8780f4bc3e53b3bfed2f))
705 |
706 |
707 | ## v5.17.3 (2025-10-02)
708 |
709 | ### Chores
710 |
711 | - Sync version to config.py [skip ci]
712 | ([`722f6f8`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/722f6f86ae228206ce0094d109a3b20499d4e11a))
713 |
714 |
715 | ## v5.17.2 (2025-10-02)
716 |
717 | ### Chores
718 |
719 | - Sync version to config.py [skip ci]
720 | ([`e47a7e8`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/e47a7e89d5bfad0bb0150cb3207f1a37dc91b170))
721 |
722 |
723 | ## v5.17.1 (2025-10-02)
724 |
725 | ### Bug Fixes
726 |
727 | - Baseclass should return MODEL_CAPABILITIES
728 | ([`82a03ce`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/82a03ce63f28fece17bfc1d70bdb75aadec4c6bb))
729 |
730 | ### Chores
731 |
732 | - Sync version to config.py [skip ci]
733 | ([`7ce66bd`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/7ce66bd9508865cef64dc30936e86e37c1a306d0))
734 |
735 | ### Documentation
736 |
737 | - Document custom timeout values
738 | ([`218fbdf`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/218fbdf49cb90f2353f58bbaef567519dd876634))
739 |
740 | ### Refactoring
741 |
742 | - Clean temperature inference
743 | ([`9c11ecc`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/9c11ecc4bf37562aa08dc3ecfa70f380e0ead357))
744 |
745 | - Cleanup
746 | ([`6ec2033`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/6ec2033f34c74ad139036de83a34cf6d374db77b))
747 |
748 | - Cleanup provider base class; cleanup shared responsibilities; cleanup public contract
749 | ([`693b84d`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/693b84db2b87271ac809abcf02100eee7405720b))
750 |
751 | - Cleanup token counting
752 | ([`7fe9fc4`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/7fe9fc49f8e3cd92be4c45a6645d5d4ab3014091))
753 |
754 | - Code cleanup
755 | ([`bb138e2`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/bb138e2fb552f837b0f9f466027580e1feb26f7c))
756 |
757 | - Code cleanup
758 | ([`182aa62`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/182aa627dfba6c578089f83444882cdd2635a7e3))
759 |
760 | - Moved image related code out of base provider into a separate utility
761 | ([`14a35af`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/14a35afa1d25408e62b968d9846be7bffaede327))
762 |
763 | - Moved temperature method from base provider to model capabilities
764 | ([`6d237d0`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/6d237d09709f757a042baf655f47eb4ddfc078ad))
765 |
766 | - Moved temperature method from base provider to model capabilities
767 | ([`f461cb4`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/f461cb451953f882bbde096a9ecf0584deb1dde8))
768 |
769 | - Removed hard coded checks, use model capabilities instead
770 | ([`250545e`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/250545e34f8d4f8026bfebb3171f3c2bc40f4692))
771 |
772 | - Removed hook from base class, turned into helper static method instead
773 | ([`2b10adc`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/2b10adcaf2b8741f0da5de84cc3483eae742a014))
774 |
775 | - Removed method from provider, should use model capabilities instead
776 | ([`a254ff2`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/a254ff2220ba00ec30f5110c69a4841419917382))
777 |
778 | - Renaming to reflect underlying type
779 | ([`1dc25f6`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/1dc25f6c3d4cdbf01f041cc424e3b5235c23175b))
780 |
781 |
782 | ## v5.17.0 (2025-10-02)
783 |
784 | ### Bug Fixes
785 |
786 | - Use types.HttpOptions from module imports instead of local import
787 | ([`956e8a6`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/956e8a6927837f5c7f031a0db1dd0b0b5483c626))
788 |
789 | ### Chores
790 |
791 | - Sync version to config.py [skip ci]
792 | ([`0836213`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/0836213071d0037d8a6d2e64d34ab5df79b8e684))
793 |
794 | ### Code Style
795 |
796 | - Apply Black formatting to use double quotes
797 | ([`33ea896`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/33ea896c511764904bf2b6b22df823928f88a148))
798 |
799 | ### Features
800 |
801 | - Add custom Gemini endpoint support
802 | ([`462bce0`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/462bce002e2141b342260969588e69f55f8bb46a))
803 |
804 | ### Refactoring
805 |
806 | - Simplify Gemini provider initialization using kwargs dict
807 | ([`023940b`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/023940be3e38a7eedbc8bf8404a4a5afc50f8398))
808 |
809 |
810 | ## v5.16.0 (2025-10-01)
811 |
812 | ### Bug Fixes
813 |
814 | - Resolve logging timing and import organization issues
815 | ([`d34c299`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/d34c299f02a233af4f17bdcc848219bf07799723))
816 |
817 | ### Chores
818 |
819 | - Sync version to config.py [skip ci]
820 | ([`b6c4bca`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/b6c4bca158e4cee1ae4abd08b7e55216ebffba2d))
821 |
822 | ### Code Style
823 |
824 | - Fix ruff import sorting issue
825 | ([`4493a69`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/4493a693332e0532d04ad3634de2a2f5b1249b64))
826 |
827 | ### Features
828 |
829 | - Add configurable environment variable override system
830 | ([`93ce698`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/93ce6987b6e7d8678ffa5ac51f5106a7a21ce67b))
831 |
832 |
833 | ## v5.15.0 (2025-10-01)
834 |
835 | ### Chores
836 |
837 | - Sync version to config.py [skip ci]
838 | ([`b0fe956`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/b0fe956f8a50240507e0fc911f0800634c15e9f7))
839 |
840 | ### Features
841 |
842 | - Depending on the number of tools in use, this change should save ~50% of overall tokens used.
843 | fixes https://github.com/BeehiveInnovations/zen-mcp-server/issues/255 but also refactored
844 | individual tools to instead encourage the agent to use the listmodels tool if needed.
845 | ([`d9449c7`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/d9449c7bb607caff3f0454f210ddfc36256c738a))
846 |
847 | ### Performance Improvements
848 |
849 | - Tweaks to schema descriptions, aiming to reduce token usage without performance degradation
850 | ([`cc8a4df`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/cc8a4dfd21b6f3dae4972a833b619e53c964693b))
851 |
852 | ### Refactoring
853 |
854 | - Trimmed some prompts
855 | ([`f69ff03`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/f69ff03c4d10e606a1dfed2a167f3ba2e2236ba8))
856 |
857 |
858 | ## v5.14.1 (2025-10-01)
859 |
860 | ### Bug Fixes
861 |
862 | - Https://github.com/BeehiveInnovations/zen-mcp-server/issues/258
863 | ([`696b45f`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/696b45f25e80faccb67034254cf9a8fc4c643dbd))
864 |
865 | ### Chores
866 |
867 | - Sync version to config.py [skip ci]
868 | ([`692016c`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/692016c6205ed0a0c3d9e830482d88231aca2e31))
869 |
870 |
871 | ## v5.14.0 (2025-10-01)
872 |
873 | ### Chores
874 |
875 | - Sync version to config.py [skip ci]
876 | ([`c0f822f`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/c0f822ffa23292d668f7b5dd3cb62e3f23fb29af))
877 |
878 | ### Features
879 |
880 | - Add Claude Sonnet 4.5 and update alias configuration
881 | ([`95c4822`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/95c4822af2dc55f59c0e4ed9454673d6ca964731))
882 |
883 | ### Testing
884 |
885 | - Update tests to match new Claude Sonnet 4.5 alias configuration
886 | ([`7efb409`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/7efb4094d4eb7db006340d3d9240b9113ac25cd3))
887 |
888 |
889 | ## v5.13.0 (2025-10-01)
890 |
891 | ### Bug Fixes
892 |
893 | - Add sonnet alias for Claude Sonnet 4.1 to match opus/haiku pattern
894 | ([`dc96344`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/dc96344db043e087ee4f8bf264a79c51dc2e0b7a))
895 |
896 | - Missing "optenai/" in name
897 | ([`7371ed6`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/7371ed6487b7d90a1b225a67dca2a38c1a52f2ad))
898 |
899 | ### Chores
900 |
901 | - Sync version to config.py [skip ci]
902 | ([`b8479fc`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/b8479fc638083d6caa4bad6205e3d3fcab830aca))
903 |
904 | ### Features
905 |
906 | - Add comprehensive GPT-5 series model support
907 | ([`4930824`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/493082405237e66a2f033481a5f8bf8293b0d553))
908 |
909 |
910 | ## v5.12.1 (2025-10-01)
911 |
912 | ### Bug Fixes
913 |
914 | - Resolve consensus tool model_context parameter missing issue
915 | ([`9044b63`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/9044b63809113047fe678d659e4fcd175f58e87a))
916 |
917 | ### Chores
918 |
919 | - Sync version to config.py [skip ci]
920 | ([`e3ebf4e`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/e3ebf4e94eba63acdc4df5a0b0493e44e3343dd1))
921 |
922 | ### Code Style
923 |
924 | - Fix trailing whitespace in consensus.py
925 | ([`0760b31`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/0760b31f8a6d03c4bea3fd2a94dfbbfab0ad5079))
926 |
927 | ### Refactoring
928 |
929 | - Optimize ModelContext creation in consensus tool
930 | ([`30a8952`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/30a8952fbccd22bebebd14eb2c8005404b79bcd6))
931 |
932 |
933 | ## v5.12.0 (2025-10-01)
934 |
935 | ### Bug Fixes
936 |
937 | - Removed use_websearch; this parameter was confusing Codex. It started using this to prompt the
938 | external model to perform searches! web-search is enabled by Claude / Codex etc by default and the
939 | external agent can ask claude to search on its behalf.
940 | ([`cff6d89`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/cff6d8998f64b73265c4e31b2352462d6afe377f))
941 |
942 | ### Chores
943 |
944 | - Sync version to config.py [skip ci]
945 | ([`28cabe0`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/28cabe0833661b0bab56d4227781ee2da332b00c))
946 |
947 | ### Features
948 |
949 | - Implement semantic cassette matching for o3 models
950 | ([`70fa088`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/70fa088c32ac4e6153d5e7b30a3e32022be2f908))
951 |
952 |
953 | ## v5.11.2 (2025-10-01)
954 |
955 | ### Chores
956 |
957 | - Sync version to config.py [skip ci]
958 | ([`4d6f1b4`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/4d6f1b41005dee428c955e33f04f8f9f6259e662))
959 |
960 |
961 | ## v5.11.1 (2025-10-01)
962 |
963 | ### Bug Fixes
964 |
965 | - Remove duplicate OpenAI models from listmodels output
966 | ([`c29e762`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/c29e7623ace257eb45396cdf8c19e1659e29edb9))
967 |
968 | ### Chores
969 |
970 | - Sync version to config.py [skip ci]
971 | ([`1209064`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/12090646ee83f2368311d595d87ae947e46ddacd))
972 |
973 | ### Testing
974 |
975 | - Update OpenAI provider alias tests to match new format
976 | ([`d13700c`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/d13700c14c7ee3d092302837cb1726d17bab1ab8))
977 |
978 |
979 | ## v5.11.0 (2025-08-26)
980 |
981 | ### Chores
982 |
983 | - Sync version to config.py [skip ci]
984 | ([`9735469`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/973546990f2c45afa93f1aa6de33ff461ecf1a83))
985 |
986 | ### Features
987 |
988 | - Codex CLI support
989 | ([`ce56d16`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/ce56d16240ddcc476145a512561efe5c66438f0d))
990 |
991 |
992 | ## v5.10.3 (2025-08-24)
993 |
994 | ### Bug Fixes
995 |
996 | - Address test failures and PR feedback
997 | ([`6bd9d67`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/6bd9d6709acfb584ab30a0a4d6891cabdb6d3ccf))
998 |
999 | - Resolve temperature handling issues for O3/custom models
1000 | ([#245](https://github.com/BeehiveInnovations/zen-mcp-server/pull/245),
1001 | [`3b4fd88`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/3b4fd88d7e9a3f09fea616a10cb3e9d6c1a0d63b))
1002 |
1003 | ### Chores
1004 |
1005 | - Sync version to config.py [skip ci]
1006 | ([`d6e6808`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/d6e6808be525192ab8388c0f01bc1bbd016fc23a))
1007 |
1008 |
1009 | ## v5.10.2 (2025-08-24)
1010 |
1011 | ### Bug Fixes
1012 |
1013 | - Another fix for https://github.com/BeehiveInnovations/zen-mcp-server/issues/251
1014 | ([`a07036e`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/a07036e6805042895109c00f921c58a09caaa319))
1015 |
1016 | ### Chores
1017 |
1018 | - Sync version to config.py [skip ci]
1019 | ([`9da5c37`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/9da5c37809cbde19d0c7ffed273ae93ca883a016))
1020 |
1021 |
1022 | ## v5.10.0 (2025-08-22)
1023 |
1024 | ### Chores
1025 |
1026 | - Sync version to config.py [skip ci]
1027 | ([`1254205`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/12542054a214022d3f515e53367f5bf3a77fb289))
1028 |
1029 | ### Features
1030 |
1031 | - Refactored and tweaked model descriptions / schema to use fewer tokens at launch (average
1032 | reduction per field description: 60-80%) without sacrificing tool effectiveness
1033 | ([`4b202f5`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/4b202f5d1d24cea1394adab26a976188f847bd09))
1034 |
1035 |
1036 | ## v5.9.0 (2025-08-21)
1037 |
1038 | ### Documentation
1039 |
1040 | - Update instructions for precommit
1041 | ([`90821b5`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/90821b51ff653475d9fb1bc70b57951d963e8841))
1042 |
1043 | ### Features
1044 |
1045 | - Refactored and improved codereview in line with precommit. Reviews are now either external
1046 | (default) or internal. Takes away anxiety and loss of tokens when Claude incorrectly decides to be
1047 | 'confident' about its own changes and bungle things up.
1048 | ([`80d21e5`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/80d21e57c0246762c0a306ede5b93d6aeb2315d8))
1049 |
1050 | ### Refactoring
1051 |
1052 | - Minor prompt tweaks
1053 | ([`d30c212`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/d30c212029c05b767d99b5391c1dd4cee78ef336))
1054 |
1055 |
1056 | ## v5.8.6 (2025-08-20)
1057 |
1058 | ### Bug Fixes
1059 |
1060 | - Escape backslashes in TOML regex pattern
1061 | ([`1c973af`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/1c973afb002650b9bbee8a831b756bef848915a1))
1062 |
1063 | - Establish version 5.8.6 and add version sync automation
1064 | ([`90a4195`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/90a419538128b54fbd30da4b8a8088ac59f8c691))
1065 |
1066 | - Restore proper version 5.8.6
1067 | ([`340b58f`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/340b58f2e790b84c3736aa96df7f6f5f2d6a13c9))
1068 |
1069 | ### Chores
1070 |
1071 | - Sync version to config.py [skip ci]
1072 | ([`4f82f65`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/4f82f6500502b7b6ba41875a560c41f6a63b683b))
1073 |
1074 |
1075 | ## v1.1.0 (2025-08-20)
1076 |
1077 | ### Features
1078 |
1079 | - Improvements to precommit
1080 | ([`2966dcf`](https://github.com/BeehiveInnovations/zen-mcp-server/commit/2966dcf2682feb7eef4073738d0c225a44ce0533))
1081 |
1082 |
1083 | ## v1.0.0 (2025-08-20)
1084 |
1085 | - Initial Release
1086 |
```
--------------------------------------------------------------------------------
/tools/consensus.py:
--------------------------------------------------------------------------------
```python
1 | """
2 | Consensus tool - Step-by-step multi-model consensus with expert analysis
3 |
4 | This tool provides a structured workflow for gathering consensus from multiple models.
5 | It guides the CLI agent through systematic steps where the CLI agent first provides its own analysis,
6 | then consults each requested model one by one, and finally synthesizes all perspectives.
7 |
8 | Key features:
9 | - Step-by-step consensus workflow with progress tracking
10 | - The CLI agent's initial neutral analysis followed by model-specific consultations
11 | - Context-aware file embedding
12 | - Support for stance-based analysis (for/against/neutral)
13 | - Final synthesis combining all perspectives
14 | """
15 |
16 | from __future__ import annotations
17 |
18 | import json
19 | import logging
20 | from typing import TYPE_CHECKING, Any
21 |
22 | from pydantic import Field, model_validator
23 |
24 | if TYPE_CHECKING:
25 | from tools.models import ToolModelCategory
26 |
27 | from mcp.types import TextContent
28 |
29 | from config import TEMPERATURE_ANALYTICAL
30 | from systemprompts import CONSENSUS_PROMPT
31 | from tools.shared.base_models import ConsolidatedFindings, WorkflowRequest
32 | from utils.conversation_memory import MAX_CONVERSATION_TURNS, create_thread, get_thread
33 |
34 | from .workflow.base import WorkflowTool
35 |
36 | logger = logging.getLogger(__name__)
37 |
38 | # Tool-specific field descriptions for consensus workflow
39 | CONSENSUS_WORKFLOW_FIELD_DESCRIPTIONS = {
40 | "step": (
41 | "Consensus prompt. Step 1: write the exact proposal/question every model will see (use 'Evaluate…', not meta commentary). "
42 | "Steps 2+: capture internal notes about the latest model response—these notes are NOT sent to other models."
43 | ),
44 | "step_number": "Current step index (starts at 1). Step 1 is your analysis; steps 2+ handle each model response.",
45 | "total_steps": "Total steps = number of models consulted plus the final synthesis step.",
46 | "next_step_required": "True if more model consultations remain; set false when ready to synthesize.",
47 | "findings": (
48 | "Step 1: your independent analysis for later synthesis (not shared with other models). Steps 2+: summarize the newest model response."
49 | ),
50 | "relevant_files": "Optional supporting files that help the consensus analysis. Must be absolute full, non-abbreviated paths.",
51 | "models": (
52 | "User-specified list of models to consult (provide at least two entries). "
53 | "Each entry may include model, stance (for/against/neutral), and stance_prompt. "
54 | "Each (model, stance) pair must be unique, e.g. [{'model':'gpt5','stance':'for'}, {'model':'pro','stance':'against'}]."
55 | ),
56 | "current_model_index": "0-based index of the next model to consult (managed internally).",
57 | "model_responses": "Internal log of responses gathered so far.",
58 | "images": "Optional absolute image paths or base64 references that add helpful visual context.",
59 | }
60 |
61 |
62 | class ConsensusRequest(WorkflowRequest):
63 | """Request model for consensus workflow steps"""
64 |
65 | # Required fields for each step
66 | step: str = Field(..., description=CONSENSUS_WORKFLOW_FIELD_DESCRIPTIONS["step"])
67 | step_number: int = Field(..., description=CONSENSUS_WORKFLOW_FIELD_DESCRIPTIONS["step_number"])
68 | total_steps: int = Field(..., description=CONSENSUS_WORKFLOW_FIELD_DESCRIPTIONS["total_steps"])
69 | next_step_required: bool = Field(..., description=CONSENSUS_WORKFLOW_FIELD_DESCRIPTIONS["next_step_required"])
70 |
71 | # Investigation tracking fields
72 | findings: str = Field(..., description=CONSENSUS_WORKFLOW_FIELD_DESCRIPTIONS["findings"])
73 | confidence: str = Field(default="exploring", exclude=True, description="Not used")
74 |
75 | # Consensus-specific fields (only needed in step 1)
76 | models: list[dict] | None = Field(None, description=CONSENSUS_WORKFLOW_FIELD_DESCRIPTIONS["models"])
77 | relevant_files: list[str] | None = Field(
78 | default_factory=list,
79 | description=CONSENSUS_WORKFLOW_FIELD_DESCRIPTIONS["relevant_files"],
80 | )
81 |
82 | # Internal tracking fields
83 | current_model_index: int | None = Field(
84 | 0,
85 | description=CONSENSUS_WORKFLOW_FIELD_DESCRIPTIONS["current_model_index"],
86 | )
87 | model_responses: list[dict] | None = Field(
88 | default_factory=list,
89 | description=CONSENSUS_WORKFLOW_FIELD_DESCRIPTIONS["model_responses"],
90 | )
91 |
92 | # Optional images for visual debugging
93 | images: list[str] | None = Field(default=None, description=CONSENSUS_WORKFLOW_FIELD_DESCRIPTIONS["images"])
94 |
95 | # Override inherited fields to exclude them from schema
96 | temperature: float | None = Field(default=None, exclude=True)
97 | thinking_mode: str | None = Field(default=None, exclude=True)
98 |
99 | # Not used in consensus workflow
100 | files_checked: list[str] | None = Field(default_factory=list, exclude=True)
101 | relevant_context: list[str] | None = Field(default_factory=list, exclude=True)
102 | issues_found: list[dict] | None = Field(default_factory=list, exclude=True)
103 | hypothesis: str | None = Field(None, exclude=True)
104 |
105 | @model_validator(mode="after")
106 | def validate_step_one_requirements(self):
107 | """Ensure step 1 has required models field and unique model+stance combinations."""
108 | if self.step_number == 1:
109 | if not self.models:
110 | raise ValueError("Step 1 requires 'models' field to specify which models to consult")
111 |
112 | # Check for unique model + stance combinations
113 | seen_combinations = set()
114 | for model_config in self.models:
115 | model_name = model_config.get("model", "")
116 | stance = model_config.get("stance", "neutral")
117 | combination = f"{model_name}:{stance}"
118 |
119 | if combination in seen_combinations:
120 | raise ValueError(
121 | f"Duplicate model + stance combination found: {model_name} with stance '{stance}'. "
122 | f"Each model + stance combination must be unique."
123 | )
124 | seen_combinations.add(combination)
125 |
126 | return self
127 |
128 |
129 | class ConsensusTool(WorkflowTool):
130 | """
131 | Consensus workflow tool for step-by-step multi-model consensus gathering.
132 |
133 | This tool implements a structured consensus workflow where the CLI agent first provides
134 | its own neutral analysis, then consults each specified model individually,
135 | and finally synthesizes all perspectives into a unified recommendation.
136 | """
137 |
138 | def __init__(self):
139 | super().__init__()
140 | self.initial_prompt: str | None = None
141 | self.original_proposal: str | None = None # Store the original proposal separately
142 | self.models_to_consult: list[dict] = []
143 | self.accumulated_responses: list[dict] = []
144 | self._current_arguments: dict[str, Any] = {}
145 |
146 | def get_name(self) -> str:
147 | return "consensus"
148 |
149 | def get_description(self) -> str:
150 | return (
151 | "Builds multi-model consensus through systematic analysis and structured debate. "
152 | "Use for complex decisions, architectural choices, feature proposals, and technology evaluations. "
153 | "Consults multiple models with different stances to synthesize comprehensive recommendations."
154 | )
155 |
156 | def get_system_prompt(self) -> str:
157 | # For the CLI agent's initial analysis, use a neutral version of the consensus prompt
158 | return CONSENSUS_PROMPT.replace(
159 | "{stance_prompt}",
160 | """BALANCED PERSPECTIVE
161 |
162 | Provide objective analysis considering both positive and negative aspects. However, if there is overwhelming evidence
163 | that the proposal clearly leans toward being exceptionally good or particularly problematic, you MUST accurately
164 | reflect this reality. Being "balanced" means being truthful about the weight of evidence, not artificially creating
165 | 50/50 splits when the reality is 90/10.
166 |
167 | Your analysis should:
168 | - Present all significant pros and cons discovered
169 | - Weight them according to actual impact and likelihood
170 | - If evidence strongly favors one conclusion, clearly state this
171 | - Provide proportional coverage based on the strength of arguments
172 | - Help the questioner see the true balance of considerations
173 |
174 | Remember: Artificial balance that misrepresents reality is not helpful. True balance means accurate representation
175 | of the evidence, even when it strongly points in one direction.""",
176 | )
177 |
178 | def get_default_temperature(self) -> float:
179 | return TEMPERATURE_ANALYTICAL
180 |
181 | def get_model_category(self) -> ToolModelCategory:
182 | """Consensus workflow requires extended reasoning"""
183 | from tools.models import ToolModelCategory
184 |
185 | return ToolModelCategory.EXTENDED_REASONING
186 |
187 | def get_workflow_request_model(self):
188 | """Return the consensus workflow-specific request model."""
189 | return ConsensusRequest
190 |
191 | def get_input_schema(self) -> dict[str, Any]:
192 | """Generate input schema for consensus workflow."""
193 | from .workflow.schema_builders import WorkflowSchemaBuilder
194 |
195 | # Consensus tool-specific field definitions
196 | consensus_field_overrides = {
197 | # Override standard workflow fields that need consensus-specific descriptions
198 | "step": {
199 | "type": "string",
200 | "description": CONSENSUS_WORKFLOW_FIELD_DESCRIPTIONS["step"],
201 | },
202 | "step_number": {
203 | "type": "integer",
204 | "minimum": 1,
205 | "description": CONSENSUS_WORKFLOW_FIELD_DESCRIPTIONS["step_number"],
206 | },
207 | "total_steps": {
208 | "type": "integer",
209 | "minimum": 1,
210 | "description": CONSENSUS_WORKFLOW_FIELD_DESCRIPTIONS["total_steps"],
211 | },
212 | "next_step_required": {
213 | "type": "boolean",
214 | "description": CONSENSUS_WORKFLOW_FIELD_DESCRIPTIONS["next_step_required"],
215 | },
216 | "findings": {
217 | "type": "string",
218 | "description": CONSENSUS_WORKFLOW_FIELD_DESCRIPTIONS["findings"],
219 | },
220 | "relevant_files": {
221 | "type": "array",
222 | "items": {"type": "string"},
223 | "description": CONSENSUS_WORKFLOW_FIELD_DESCRIPTIONS["relevant_files"],
224 | },
225 | # consensus-specific fields (not in base workflow)
226 | "models": {
227 | "type": "array",
228 | "items": {
229 | "type": "object",
230 | "properties": {
231 | "model": {"type": "string"},
232 | "stance": {"type": "string", "enum": ["for", "against", "neutral"], "default": "neutral"},
233 | "stance_prompt": {"type": "string"},
234 | },
235 | "required": ["model"],
236 | },
237 | "description": (
238 | "User-specified roster of models to consult (provide at least two entries). "
239 | + CONSENSUS_WORKFLOW_FIELD_DESCRIPTIONS["models"]
240 | ),
241 | "minItems": 2,
242 | },
243 | "current_model_index": {
244 | "type": "integer",
245 | "minimum": 0,
246 | "description": CONSENSUS_WORKFLOW_FIELD_DESCRIPTIONS["current_model_index"],
247 | },
248 | "model_responses": {
249 | "type": "array",
250 | "items": {"type": "object"},
251 | "description": CONSENSUS_WORKFLOW_FIELD_DESCRIPTIONS["model_responses"],
252 | },
253 | "images": {
254 | "type": "array",
255 | "items": {"type": "string"},
256 | "description": CONSENSUS_WORKFLOW_FIELD_DESCRIPTIONS["images"],
257 | },
258 | }
259 |
260 | # Provide guidance on available models similar to single-model tools
261 | model_description = (
262 | "When the user names a model, you MUST use that exact value or report the "
263 | "provider error—never swap in another option. Use the `listmodels` tool for the full roster."
264 | )
265 |
266 | summaries, total, restricted = self._get_ranked_model_summaries()
267 | remainder = max(0, total - len(summaries))
268 | if summaries:
269 | label = "Allowed models" if restricted else "Top models"
270 | top_line = "; ".join(summaries)
271 | if remainder > 0:
272 | top_line = f"{label}: {top_line}; +{remainder} more via `listmodels`."
273 | else:
274 | top_line = f"{label}: {top_line}."
275 | model_description = f"{model_description} {top_line}"
276 | else:
277 | model_description = (
278 | f"{model_description} No models detected—configure provider credentials or use the `listmodels` tool "
279 | "to inspect availability."
280 | )
281 |
282 | restriction_note = self._get_restriction_note()
283 | if restriction_note and (remainder > 0 or not summaries):
284 | model_description = f"{model_description} {restriction_note}."
285 |
286 | existing_models_desc = consensus_field_overrides["models"]["description"]
287 | consensus_field_overrides["models"]["description"] = f"{existing_models_desc} {model_description}"
288 |
289 | # Define excluded fields for consensus workflow
290 | excluded_workflow_fields = [
291 | "files_checked", # Not used in consensus workflow
292 | "relevant_context", # Not used in consensus workflow
293 | "issues_found", # Not used in consensus workflow
294 | "hypothesis", # Not used in consensus workflow
295 | "confidence", # Not used in consensus workflow
296 | ]
297 |
298 | excluded_common_fields = [
299 | "model", # Consensus uses 'models' field instead
300 | "temperature", # Not used in consensus workflow
301 | "thinking_mode", # Not used in consensus workflow
302 | ]
303 |
304 | requires_model = self.requires_model()
305 | model_field_schema = self.get_model_field_schema() if requires_model else None
306 | auto_mode = self.is_effective_auto_mode() if requires_model else False
307 |
308 | return WorkflowSchemaBuilder.build_schema(
309 | tool_specific_fields=consensus_field_overrides,
310 | model_field_schema=model_field_schema,
311 | auto_mode=auto_mode,
312 | tool_name=self.get_name(),
313 | excluded_workflow_fields=excluded_workflow_fields,
314 | excluded_common_fields=excluded_common_fields,
315 | require_model=requires_model,
316 | )
317 |
318 | def get_required_actions(
319 | self, step_number: int, confidence: str, findings: str, total_steps: int, request=None
320 | ) -> list[str]: # noqa: ARG002
321 | """Define required actions for each consensus phase.
322 |
323 | Now includes request parameter for continuation-aware decisions.
324 | Note: confidence parameter is kept for compatibility with base class but not used.
325 | """
326 | if step_number == 1:
327 | # CLI Agent's initial analysis
328 | return [
329 | "You've provided your initial analysis. The tool will now consult other models.",
330 | "Wait for the next step to receive the first model's response.",
331 | ]
332 | elif step_number < total_steps - 1:
333 | # Processing individual model responses
334 | return [
335 | "Review the model response provided in this step",
336 | "Note key agreements and disagreements with previous analyses",
337 | "Wait for the next model's response",
338 | ]
339 | else:
340 | # Ready for final synthesis
341 | return [
342 | "All models have been consulted",
343 | "Synthesize all perspectives into a comprehensive recommendation",
344 | "Identify key points of agreement and disagreement",
345 | "Provide clear, actionable guidance based on the consensus",
346 | ]
347 |
348 | def should_call_expert_analysis(self, consolidated_findings, request=None) -> bool:
349 | """Consensus workflow doesn't use traditional expert analysis - it consults models step by step."""
350 | return False
351 |
352 | def prepare_expert_analysis_context(self, consolidated_findings) -> str:
353 | """Not used in consensus workflow."""
354 | return ""
355 |
356 | def requires_expert_analysis(self) -> bool:
357 | """Consensus workflow handles its own model consultations."""
358 | return False
359 |
360 | def requires_model(self) -> bool:
361 | """
362 | Consensus tool doesn't require model resolution at the MCP boundary.
363 |
364 | Uses it's own set of models
365 |
366 | Returns:
367 | bool: False
368 | """
369 | return False
370 |
371 | # Hook method overrides for consensus-specific behavior
372 |
373 | def prepare_step_data(self, request) -> dict:
374 | """Prepare consensus-specific step data."""
375 | step_data = {
376 | "step": request.step,
377 | "step_number": request.step_number,
378 | "findings": request.findings,
379 | "files_checked": [], # Not used
380 | "relevant_files": request.relevant_files or [],
381 | "relevant_context": [], # Not used
382 | "issues_found": [], # Not used
383 | "confidence": "exploring", # Not used, kept for compatibility
384 | "hypothesis": None, # Not used
385 | "images": request.images or [], # Now used for visual context
386 | }
387 | return step_data
388 |
389 | async def handle_work_completion(self, response_data: dict, request, arguments: dict) -> dict: # noqa: ARG002
390 | """Handle consensus workflow completion - no expert analysis, just final synthesis."""
391 | response_data["consensus_complete"] = True
392 | response_data["status"] = "consensus_workflow_complete"
393 |
394 | # Prepare final synthesis data
395 | response_data["complete_consensus"] = {
396 | "initial_prompt": self.original_proposal if self.original_proposal else self.initial_prompt,
397 | "models_consulted": [m["model"] + ":" + m.get("stance", "neutral") for m in self.accumulated_responses],
398 | "total_responses": len(self.accumulated_responses),
399 | "consensus_confidence": "high", # Consensus complete
400 | }
401 |
402 | response_data["next_steps"] = (
403 | "CONSENSUS GATHERING IS COMPLETE. You MUST now synthesize all perspectives and present:\n"
404 | "1. Key points of AGREEMENT across models\n"
405 | "2. Key points of DISAGREEMENT and why they differ\n"
406 | "3. Your final consolidated recommendation\n"
407 | "4. Specific, actionable next steps for implementation\n"
408 | "5. Critical risks or concerns that must be addressed"
409 | )
410 |
411 | return response_data
412 |
413 | def handle_work_continuation(self, response_data: dict, request) -> dict:
414 | """Handle continuation between consensus steps."""
415 | current_idx = request.current_model_index or 0
416 |
417 | if request.step_number == 1:
418 | # After CLI Agent's initial analysis, prepare to consult first model
419 | response_data["status"] = "consulting_models"
420 | response_data["next_model"] = self.models_to_consult[0] if self.models_to_consult else None
421 | response_data["next_steps"] = (
422 | "Your initial analysis is complete. The tool will now consult the specified models."
423 | )
424 | elif current_idx < len(self.models_to_consult):
425 | next_model = self.models_to_consult[current_idx]
426 | response_data["status"] = "consulting_next_model"
427 | response_data["next_model"] = next_model
428 | response_data["models_remaining"] = len(self.models_to_consult) - current_idx
429 | response_data["next_steps"] = f"Model consultation in progress. Next: {next_model['model']}"
430 | else:
431 | response_data["status"] = "ready_for_synthesis"
432 | response_data["next_steps"] = "All models consulted. Ready for final synthesis."
433 |
434 | return response_data
435 |
436 | async def execute_workflow(self, arguments: dict[str, Any]) -> list:
437 | """Override execute_workflow to handle model consultations between steps."""
438 |
439 | # Store arguments
440 | self._current_arguments = arguments
441 |
442 | # Validate request
443 | request = self.get_workflow_request_model()(**arguments)
444 |
445 | # Resolve existing continuation_id or create a new one on first step
446 | continuation_id = request.continuation_id
447 |
448 | if request.step_number == 1:
449 | if not continuation_id:
450 | clean_args = {k: v for k, v in arguments.items() if k not in ["_model_context", "_resolved_model_name"]}
451 | continuation_id = create_thread(self.get_name(), clean_args)
452 | request.continuation_id = continuation_id
453 | arguments["continuation_id"] = continuation_id
454 | self.work_history = []
455 | self.consolidated_findings = ConsolidatedFindings()
456 |
457 | # Store the original proposal from step 1 - this is what all models should see
458 | self.store_initial_issue(request.step)
459 | self.initial_request = request.step
460 | self.models_to_consult = request.models or []
461 | self.accumulated_responses = []
462 | # Set total steps: len(models) (each step includes consultation + response)
463 | request.total_steps = len(self.models_to_consult)
464 |
465 | # For all steps (1 through total_steps), consult the corresponding model
466 | if request.step_number <= request.total_steps:
467 | # Calculate which model to consult for this step
468 | model_idx = request.step_number - 1 # 0-based index
469 |
470 | if model_idx < len(self.models_to_consult):
471 | # Track workflow state for conversation memory
472 | step_data = self.prepare_step_data(request)
473 | self.work_history.append(step_data)
474 | self._update_consolidated_findings(step_data)
475 |
476 | # Consult the model for this step
477 | model_response = await self._consult_model(self.models_to_consult[model_idx], request)
478 |
479 | # Add to accumulated responses
480 | self.accumulated_responses.append(model_response)
481 |
482 | # Include the model response in the step data
483 | response_data = {
484 | "status": "model_consulted",
485 | "step_number": request.step_number,
486 | "total_steps": request.total_steps,
487 | "model_consulted": model_response["model"],
488 | "model_stance": model_response.get("stance", "neutral"),
489 | "model_response": model_response,
490 | "current_model_index": model_idx + 1,
491 | "next_step_required": request.step_number < request.total_steps,
492 | }
493 |
494 | # Add CLAI Agent's analysis to step 1
495 | if request.step_number == 1:
496 | response_data["agent_analysis"] = {
497 | "initial_analysis": request.step,
498 | "findings": request.findings,
499 | }
500 | response_data["status"] = "analysis_and_first_model_consulted"
501 |
502 | # Check if this is the final step
503 | if request.step_number == request.total_steps:
504 | response_data["status"] = "consensus_workflow_complete"
505 | response_data["consensus_complete"] = True
506 | response_data["complete_consensus"] = {
507 | "initial_prompt": self.original_proposal if self.original_proposal else self.initial_prompt,
508 | "models_consulted": [
509 | f"{m['model']}:{m.get('stance', 'neutral')}" for m in self.accumulated_responses
510 | ],
511 | "total_responses": len(self.accumulated_responses),
512 | "consensus_confidence": "high",
513 | }
514 | response_data["next_steps"] = (
515 | "CONSENSUS GATHERING IS COMPLETE. Synthesize all perspectives and present:\n"
516 | "1. Key points of AGREEMENT across models\n"
517 | "2. Key points of DISAGREEMENT and why they differ\n"
518 | "3. Your final consolidated recommendation\n"
519 | "4. Specific, actionable next steps for implementation\n"
520 | "5. Critical risks or concerns that must be addressed"
521 | )
522 | else:
523 | response_data["next_steps"] = (
524 | f"Model {model_response['model']} has provided its {model_response.get('stance', 'neutral')} "
525 | f"perspective. Please analyze this response and call {self.get_name()} again with:\n"
526 | f"- step_number: {request.step_number + 1}\n"
527 | f"- findings: Summarize key points from this model's response"
528 | )
529 |
530 | # Add continuation information and workflow customization
531 | response_data = self.customize_workflow_response(response_data, request)
532 |
533 | # Ensure consensus-specific metadata is attached
534 | self._add_workflow_metadata(response_data, arguments)
535 |
536 | if continuation_id:
537 | self.store_conversation_turn(continuation_id, response_data, request)
538 | continuation_offer = self._build_continuation_offer(continuation_id)
539 | if continuation_offer:
540 | response_data["continuation_offer"] = continuation_offer
541 |
542 | return [TextContent(type="text", text=json.dumps(response_data, indent=2, ensure_ascii=False))]
543 |
544 | # Otherwise, use standard workflow execution
545 | return await super().execute_workflow(arguments)
546 |
547 | def _build_continuation_offer(self, continuation_id: str) -> dict[str, Any] | None:
548 | """Create a continuation offer without exposing prior model responses."""
549 | try:
550 | from tools.models import ContinuationOffer
551 |
552 | thread = get_thread(continuation_id)
553 | if thread and thread.turns:
554 | remaining_turns = max(0, MAX_CONVERSATION_TURNS - len(thread.turns))
555 | else:
556 | remaining_turns = MAX_CONVERSATION_TURNS - 1
557 |
558 | # Provide a neutral note specific to consensus workflow
559 | note = (
560 | f"Consensus workflow can continue for {remaining_turns} more exchanges."
561 | if remaining_turns > 0
562 | else "Consensus workflow continuation limit reached."
563 | )
564 |
565 | continuation_offer = ContinuationOffer(
566 | continuation_id=continuation_id,
567 | note=note,
568 | remaining_turns=remaining_turns,
569 | )
570 | return continuation_offer.model_dump()
571 | except Exception:
572 | return None
573 |
574 | async def _consult_model(self, model_config: dict, request) -> dict:
575 | """Consult a single model and return its response."""
576 | try:
577 | # Import and create ModelContext once at the beginning
578 | from utils.model_context import ModelContext
579 |
580 | # Get the provider for this model
581 | model_name = model_config["model"]
582 | provider = self.get_model_provider(model_name)
583 |
584 | # Create model context once and reuse for both file processing and temperature validation
585 | model_context = ModelContext(model_name=model_name)
586 |
587 | # Prepare the prompt with any relevant files
588 | # Use continuation_id=None for blinded consensus - each model should only see
589 | # original prompt + files, not conversation history or other model responses
590 | # CRITICAL: Use the original proposal from step 1, NOT what's in request.step for steps 2+!
591 | # Steps 2+ contain summaries/notes that must NEVER be sent to other models
592 | prompt = self.original_proposal if self.original_proposal else self.initial_prompt
593 | if request.relevant_files:
594 | file_content, _ = self._prepare_file_content_for_prompt(
595 | request.relevant_files,
596 | None, # Use None instead of request.continuation_id for blinded consensus
597 | "Context files",
598 | model_context=model_context,
599 | )
600 | if file_content:
601 | prompt = f"{prompt}\n\n=== CONTEXT FILES ===\n{file_content}\n=== END CONTEXT ==="
602 |
603 | # Get stance-specific system prompt
604 | stance = model_config.get("stance", "neutral")
605 | stance_prompt = model_config.get("stance_prompt")
606 | system_prompt = self._get_stance_enhanced_prompt(stance, stance_prompt)
607 |
608 | # Validate temperature against model constraints (respects supports_temperature)
609 | validated_temperature, temp_warnings = self.validate_and_correct_temperature(
610 | self.get_default_temperature(), model_context
611 | )
612 |
613 | # Log any temperature corrections
614 | for warning in temp_warnings:
615 | logger.warning(warning)
616 |
617 | # Call the model with validated temperature
618 | response = provider.generate_content(
619 | prompt=prompt,
620 | model_name=model_name,
621 | system_prompt=system_prompt,
622 | temperature=validated_temperature,
623 | thinking_mode="medium",
624 | images=request.images if request.images else None,
625 | )
626 |
627 | return {
628 | "model": model_name,
629 | "stance": stance,
630 | "status": "success",
631 | "verdict": response.content,
632 | "metadata": {
633 | "provider": provider.get_provider_type().value,
634 | "model_name": model_name,
635 | },
636 | }
637 |
638 | except Exception as e:
639 | logger.exception("Error consulting model %s", model_config)
640 | return {
641 | "model": model_config.get("model", "unknown"),
642 | "stance": model_config.get("stance", "neutral"),
643 | "status": "error",
644 | "error": str(e),
645 | }
646 |
647 | def _get_stance_enhanced_prompt(self, stance: str, custom_stance_prompt: str | None = None) -> str:
648 | """Get the system prompt with stance injection."""
649 | base_prompt = CONSENSUS_PROMPT
650 |
651 | if custom_stance_prompt:
652 | return base_prompt.replace("{stance_prompt}", custom_stance_prompt)
653 |
654 | stance_prompts = {
655 | "for": """SUPPORTIVE PERSPECTIVE WITH INTEGRITY
656 |
657 | You are tasked with advocating FOR this proposal, but with CRITICAL GUARDRAILS:
658 |
659 | MANDATORY ETHICAL CONSTRAINTS:
660 | - This is NOT a debate for entertainment. You MUST act in good faith and in the best interest of the questioner
661 | - You MUST think deeply about whether supporting this idea is safe, sound, and passes essential requirements
662 | - You MUST be direct and unequivocal in saying "this is a bad idea" when it truly is
663 | - There must be at least ONE COMPELLING reason to be optimistic, otherwise DO NOT support it
664 |
665 | WHEN TO REFUSE SUPPORT (MUST OVERRIDE STANCE):
666 | - If the idea is fundamentally harmful to users, project, or stakeholders
667 | - If implementation would violate security, privacy, or ethical standards
668 | - If the proposal is technically infeasible within realistic constraints
669 | - If costs/risks dramatically outweigh any potential benefits
670 |
671 | YOUR SUPPORTIVE ANALYSIS SHOULD:
672 | - Identify genuine strengths and opportunities
673 | - Propose solutions to overcome legitimate challenges
674 | - Highlight synergies with existing systems
675 | - Suggest optimizations that enhance value
676 | - Present realistic implementation pathways
677 |
678 | Remember: Being "for" means finding the BEST possible version of the idea IF it has merit, not blindly supporting bad ideas.""",
679 | "against": """CRITICAL PERSPECTIVE WITH RESPONSIBILITY
680 |
681 | You are tasked with critiquing this proposal, but with ESSENTIAL BOUNDARIES:
682 |
683 | MANDATORY FAIRNESS CONSTRAINTS:
684 | - You MUST NOT oppose genuinely excellent, common-sense ideas just to be contrarian
685 | - You MUST acknowledge when a proposal is fundamentally sound and well-conceived
686 | - You CANNOT give harmful advice or recommend against beneficial changes
687 | - If the idea is outstanding, say so clearly while offering constructive refinements
688 |
689 | WHEN TO MODERATE CRITICISM (MUST OVERRIDE STANCE):
690 | - If the proposal addresses critical user needs effectively
691 | - If it follows established best practices with good reason
692 | - If benefits clearly and substantially outweigh risks
693 | - If it's the obvious right solution to the problem
694 |
695 | YOUR CRITICAL ANALYSIS SHOULD:
696 | - Identify legitimate risks and failure modes
697 | - Point out overlooked complexities
698 | - Suggest more efficient alternatives
699 | - Highlight potential negative consequences
700 | - Question assumptions that may be flawed
701 |
702 | Remember: Being "against" means rigorous scrutiny to ensure quality, not undermining good ideas that deserve support.""",
703 | "neutral": """BALANCED PERSPECTIVE
704 |
705 | Provide objective analysis considering both positive and negative aspects. However, if there is overwhelming evidence
706 | that the proposal clearly leans toward being exceptionally good or particularly problematic, you MUST accurately
707 | reflect this reality. Being "balanced" means being truthful about the weight of evidence, not artificially creating
708 | 50/50 splits when the reality is 90/10.
709 |
710 | Your analysis should:
711 | - Present all significant pros and cons discovered
712 | - Weight them according to actual impact and likelihood
713 | - If evidence strongly favors one conclusion, clearly state this
714 | - Provide proportional coverage based on the strength of arguments
715 | - Help the questioner see the true balance of considerations
716 |
717 | Remember: Artificial balance that misrepresents reality is not helpful. True balance means accurate representation
718 | of the evidence, even when it strongly points in one direction.""",
719 | }
720 |
721 | stance_prompt = stance_prompts.get(stance, stance_prompts["neutral"])
722 | return base_prompt.replace("{stance_prompt}", stance_prompt)
723 |
724 | def customize_workflow_response(self, response_data: dict, request) -> dict:
725 | """Customize response for consensus workflow."""
726 | # Store model responses in the response for tracking
727 | if self.accumulated_responses:
728 | response_data["accumulated_responses"] = self.accumulated_responses
729 |
730 | # Add consensus-specific fields
731 | if request.step_number == 1:
732 | response_data["consensus_workflow_status"] = "initial_analysis_complete"
733 | elif request.step_number < request.total_steps - 1:
734 | response_data["consensus_workflow_status"] = "consulting_models"
735 | else:
736 | response_data["consensus_workflow_status"] = "ready_for_synthesis"
737 |
738 | # Customize metadata for consensus workflow
739 | self._customize_consensus_metadata(response_data, request)
740 |
741 | return response_data
742 |
743 | def _customize_consensus_metadata(self, response_data: dict, request) -> None:
744 | """
745 | Customize metadata for consensus workflow to accurately reflect multi-model nature.
746 |
747 | The default workflow metadata shows the model running Agent's analysis steps,
748 | but consensus is a multi-model tool that consults different models. We need
749 | to provide accurate metadata that reflects this.
750 | """
751 | if "metadata" not in response_data:
752 | response_data["metadata"] = {}
753 |
754 | metadata = response_data["metadata"]
755 |
756 | # Always preserve tool_name
757 | metadata["tool_name"] = self.get_name()
758 |
759 | if request.step_number == request.total_steps:
760 | # Final step - show comprehensive consensus metadata
761 | models_consulted = []
762 | if self.models_to_consult:
763 | models_consulted = [f"{m['model']}:{m.get('stance', 'neutral')}" for m in self.models_to_consult]
764 |
765 | metadata.update(
766 | {
767 | "workflow_type": "multi_model_consensus",
768 | "models_consulted": models_consulted,
769 | "consensus_complete": True,
770 | "total_models": len(self.models_to_consult) if self.models_to_consult else 0,
771 | }
772 | )
773 |
774 | # Remove the misleading single model metadata
775 | metadata.pop("model_used", None)
776 | metadata.pop("provider_used", None)
777 |
778 | else:
779 | # Intermediate steps - show consensus workflow in progress
780 | models_to_consult = []
781 | if self.models_to_consult:
782 | models_to_consult = [f"{m['model']}:{m.get('stance', 'neutral')}" for m in self.models_to_consult]
783 |
784 | metadata.update(
785 | {
786 | "workflow_type": "multi_model_consensus",
787 | "models_to_consult": models_to_consult,
788 | "consultation_step": request.step_number,
789 | "total_consultation_steps": request.total_steps,
790 | }
791 | )
792 |
793 | # Remove the misleading single model metadata that shows Agent's execution model
794 | # instead of the models being consulted
795 | metadata.pop("model_used", None)
796 | metadata.pop("provider_used", None)
797 |
798 | def _add_workflow_metadata(self, response_data: dict, arguments: dict[str, Any]) -> None:
799 | """
800 | Override workflow metadata addition for consensus tool.
801 |
802 | The consensus tool doesn't use single model metadata because it's a multi-model
803 | workflow. Instead, we provide consensus-specific metadata that accurately
804 | reflects the models being consulted.
805 | """
806 | # Initialize metadata if not present
807 | if "metadata" not in response_data:
808 | response_data["metadata"] = {}
809 |
810 | # Add basic tool metadata
811 | response_data["metadata"]["tool_name"] = self.get_name()
812 |
813 | # The consensus-specific metadata is already added by _customize_consensus_metadata
814 | # which is called from customize_workflow_response. We don't add the standard
815 | # single-model metadata (model_used, provider_used) because it's misleading
816 | # for a multi-model consensus workflow.
817 |
818 | logger.debug(
819 | f"[CONSENSUS_METADATA] {self.get_name()}: Using consensus-specific metadata instead of single-model metadata"
820 | )
821 |
822 | def store_initial_issue(self, step_description: str):
823 | """Store initial prompt for model consultations."""
824 | self.original_proposal = step_description
825 | self.initial_prompt = step_description # Keep for backward compatibility
826 |
827 | # Required abstract methods from BaseTool
828 | def get_request_model(self):
829 | """Return the consensus workflow-specific request model."""
830 | return ConsensusRequest
831 |
832 | async def prepare_prompt(self, request) -> str: # noqa: ARG002
833 | """Not used - workflow tools use execute_workflow()."""
834 | return "" # Workflow tools use execute_workflow() directly
835 |
```
--------------------------------------------------------------------------------
/tools/precommit.py:
--------------------------------------------------------------------------------
```python
1 | """
2 | Precommit Workflow tool - Step-by-step pre-commit validation with expert analysis
3 |
4 | This tool provides a structured workflow for comprehensive pre-commit validation.
5 | It guides the CLI agent through systematic investigation steps with forced pauses between each step
6 | to ensure thorough code examination, git change analysis, and issue detection before proceeding.
7 | The tool supports finding updates and expert analysis integration.
8 |
9 | Key features:
10 | - Step-by-step pre-commit investigation workflow with progress tracking
11 | - Context-aware file embedding (references during investigation, full content for analysis)
12 | - Automatic git repository discovery and change analysis
13 | - Expert analysis integration with external models (default)
14 | - Support for multiple repositories and change types
15 | - Configurable validation type (external with expert model or internal only)
16 | """
17 |
18 | import logging
19 | from typing import TYPE_CHECKING, Any, Literal, Optional
20 |
21 | from pydantic import Field, model_validator
22 |
23 | if TYPE_CHECKING:
24 | from tools.models import ToolModelCategory
25 |
26 | from config import TEMPERATURE_ANALYTICAL
27 | from systemprompts import PRECOMMIT_PROMPT
28 | from tools.shared.base_models import WorkflowRequest
29 |
30 | from .workflow.base import WorkflowTool
31 |
32 | logger = logging.getLogger(__name__)
33 |
34 | # Tool-specific field descriptions for precommit workflow
35 | PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS = {
36 | "step": (
37 | "Step 1: outline how you'll validate the git changes. Later steps: report findings. Review diffs and impacts, use `relevant_files`, and avoid pasting large snippets."
38 | ),
39 | "step_number": "Current pre-commit step number (starts at 1).",
40 | "total_steps": (
41 | "Planned number of validation steps. External validation: use at most three (analysis → follow-ups → summary). Internal validation: a single step. Honour these limits when resuming via continuation_id."
42 | ),
43 | "next_step_required": (
44 | "True to continue with another step, False when validation is complete. "
45 | "CRITICAL: If total_steps>=3 or when `precommit_type = external`, set to True until the final step. "
46 | "When continuation_id is provided: Follow the same validation rules based on precommit_type."
47 | ),
48 | "findings": "Record git diff insights, risks, missing tests, security concerns, and positives; update previous notes as you go.",
49 | "files_checked": "Absolute paths for every file examined, including ruled-out candidates.",
50 | "relevant_files": "Absolute paths of files involved in the change or validation (code, configs, tests, docs). Must be absolute full non-abbreviated paths.",
51 | "relevant_context": "Key functions/methods touched by the change (e.g. 'Class.method', 'function_name').",
52 | "issues_found": "List issues with severity (critical/high/medium/low) plus descriptions (bugs, security, performance, coverage).",
53 | "precommit_type": "'external' (default, triggers expert model) or 'internal' (local-only validation).",
54 | "images": "Optional absolute paths to screenshots or diagrams that aid validation.",
55 | "path": "Absolute path to the repository root. Required in step 1.",
56 | "compare_to": "Optional git ref (branch/tag/commit) to diff against; falls back to staged/unstaged changes.",
57 | "include_staged": "Whether to inspect staged changes (ignored when `compare_to` is set).",
58 | "include_unstaged": "Whether to inspect unstaged changes (ignored when `compare_to` is set).",
59 | "focus_on": "Optional emphasis areas such as security, performance, or test coverage.",
60 | "severity_filter": "Lowest severity to include when reporting issues.",
61 | }
62 |
63 |
64 | class PrecommitRequest(WorkflowRequest):
65 | """Request model for precommit workflow investigation steps"""
66 |
67 | # Required fields for each investigation step
68 | step: str = Field(..., description=PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS["step"])
69 | step_number: int = Field(..., description=PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS["step_number"])
70 | total_steps: int = Field(..., description=PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS["total_steps"])
71 | next_step_required: bool = Field(..., description=PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS["next_step_required"])
72 |
73 | # Investigation tracking fields
74 | findings: str = Field(..., description=PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS["findings"])
75 | files_checked: list[str] = Field(
76 | default_factory=list, description=PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS["files_checked"]
77 | )
78 | relevant_files: list[str] = Field(
79 | default_factory=list, description=PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS["relevant_files"]
80 | )
81 | relevant_context: list[str] = Field(
82 | default_factory=list, description=PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS["relevant_context"]
83 | )
84 | issues_found: list[dict] = Field(
85 | default_factory=list, description=PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS["issues_found"]
86 | )
87 | precommit_type: Optional[Literal["external", "internal"]] = Field(
88 | "external", description=PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS["precommit_type"]
89 | )
90 |
91 | # Optional images for visual validation
92 | images: Optional[list[str]] = Field(default=None, description=PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS["images"])
93 |
94 | # Precommit-specific fields (only used in step 1 to initialize)
95 | # Required for step 1, validated in model_validator
96 | path: Optional[str] = Field(None, description=PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS["path"])
97 | compare_to: Optional[str] = Field(None, description=PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS["compare_to"])
98 | include_staged: Optional[bool] = Field(True, description=PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS["include_staged"])
99 | include_unstaged: Optional[bool] = Field(
100 | True, description=PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS["include_unstaged"]
101 | )
102 | focus_on: Optional[str] = Field(None, description=PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS["focus_on"])
103 | severity_filter: Optional[Literal["critical", "high", "medium", "low", "all"]] = Field(
104 | "all", description=PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS["severity_filter"]
105 | )
106 |
107 | # Override inherited fields to exclude them from schema (except model which needs to be available)
108 | temperature: Optional[float] = Field(default=None, exclude=True)
109 | thinking_mode: Optional[str] = Field(default=None, exclude=True)
110 |
111 | @model_validator(mode="after")
112 | def validate_step_one_requirements(self):
113 | """Ensure step 1 has required path field."""
114 | if self.step_number == 1 and not self.path:
115 | raise ValueError("Step 1 requires 'path' field to specify git repository location")
116 | return self
117 |
118 |
119 | class PrecommitTool(WorkflowTool):
120 | """
121 | Precommit workflow tool for step-by-step pre-commit validation and expert analysis.
122 |
123 | This tool implements a structured pre-commit validation workflow that guides users through
124 | methodical investigation steps, ensuring thorough change examination, issue identification,
125 | and validation before reaching conclusions. It supports complex validation scenarios including
126 | multi-repository analysis, security review, performance validation, and integration testing.
127 | """
128 |
129 | def __init__(self):
130 | super().__init__()
131 | self.initial_request = None
132 | self.git_config = {}
133 |
134 | def get_name(self) -> str:
135 | return "precommit"
136 |
137 | def get_description(self) -> str:
138 | return (
139 | "Validates git changes and repository state before committing with systematic analysis. "
140 | "Use for multi-repository validation, security review, change impact assessment, and completeness verification. "
141 | "Guides through structured investigation with expert analysis."
142 | )
143 |
144 | def get_system_prompt(self) -> str:
145 | return PRECOMMIT_PROMPT
146 |
147 | def get_default_temperature(self) -> float:
148 | return TEMPERATURE_ANALYTICAL
149 |
150 | def get_model_category(self) -> "ToolModelCategory":
151 | """Precommit requires thorough analysis and reasoning"""
152 | from tools.models import ToolModelCategory
153 |
154 | return ToolModelCategory.EXTENDED_REASONING
155 |
156 | def get_workflow_request_model(self):
157 | """Return the precommit workflow-specific request model."""
158 | return PrecommitRequest
159 |
160 | def get_input_schema(self) -> dict[str, Any]:
161 | """Generate input schema using WorkflowSchemaBuilder with precommit-specific overrides."""
162 | from .workflow.schema_builders import WorkflowSchemaBuilder
163 |
164 | # Precommit workflow-specific field overrides
165 | precommit_field_overrides = {
166 | "step": {
167 | "type": "string",
168 | "description": PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS["step"],
169 | },
170 | "step_number": {
171 | "type": "integer",
172 | "minimum": 1,
173 | "description": PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS["step_number"],
174 | },
175 | "total_steps": {
176 | "type": "integer",
177 | "minimum": 3,
178 | "description": PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS["total_steps"],
179 | },
180 | "next_step_required": {
181 | "type": "boolean",
182 | "description": PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS["next_step_required"],
183 | },
184 | "findings": {
185 | "type": "string",
186 | "description": PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS["findings"],
187 | },
188 | "files_checked": {
189 | "type": "array",
190 | "items": {"type": "string"},
191 | "description": PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS["files_checked"],
192 | },
193 | "relevant_files": {
194 | "type": "array",
195 | "items": {"type": "string"},
196 | "description": PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS["relevant_files"],
197 | },
198 | "precommit_type": {
199 | "type": "string",
200 | "enum": ["external", "internal"],
201 | "default": "external",
202 | "description": PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS["precommit_type"],
203 | },
204 | "issues_found": {
205 | "type": "array",
206 | "items": {"type": "object"},
207 | "description": PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS["issues_found"],
208 | },
209 | "images": {
210 | "type": "array",
211 | "items": {"type": "string"},
212 | "description": PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS["images"],
213 | },
214 | # Precommit-specific fields (for step 1)
215 | "path": {
216 | "type": "string",
217 | "description": PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS["path"],
218 | },
219 | "compare_to": {
220 | "type": "string",
221 | "description": PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS["compare_to"],
222 | },
223 | "include_staged": {
224 | "type": "boolean",
225 | "default": True,
226 | "description": PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS["include_staged"],
227 | },
228 | "include_unstaged": {
229 | "type": "boolean",
230 | "default": True,
231 | "description": PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS["include_unstaged"],
232 | },
233 | "focus_on": {
234 | "type": "string",
235 | "description": PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS["focus_on"],
236 | },
237 | "severity_filter": {
238 | "type": "string",
239 | "enum": ["critical", "high", "medium", "low", "all"],
240 | "default": "all",
241 | "description": PRECOMMIT_WORKFLOW_FIELD_DESCRIPTIONS["severity_filter"],
242 | },
243 | }
244 |
245 | # Use WorkflowSchemaBuilder with precommit-specific tool fields
246 | return WorkflowSchemaBuilder.build_schema(
247 | tool_specific_fields=precommit_field_overrides,
248 | model_field_schema=self.get_model_field_schema(),
249 | auto_mode=self.is_effective_auto_mode(),
250 | tool_name=self.get_name(),
251 | )
252 |
253 | def get_required_actions(
254 | self, step_number: int, confidence: str, findings: str, total_steps: int, request=None
255 | ) -> list[str]:
256 | """Define required actions for each investigation phase.
257 |
258 | Now includes request parameter for continuation-aware decisions.
259 | """
260 | # Check for continuation - fast track mode
261 | if request:
262 | continuation_id = self.get_request_continuation_id(request)
263 | precommit_type = self.get_precommit_type(request)
264 | if continuation_id and precommit_type == "external":
265 | if step_number == 1:
266 | return [
267 | "Execute git status to see all changes",
268 | "Execute git diff --cached for staged changes (exclude binary files)",
269 | "Execute git diff for unstaged changes (exclude binary files)",
270 | "List any relevant untracked files as well.",
271 | ]
272 | else:
273 | return ["Complete validation and proceed to expert analysis with changeset file"]
274 |
275 | # Extract counts for normal flow
276 | findings_count = len(findings.split("\n")) if findings else 0
277 | issues_count = self.get_consolidated_issues_count()
278 |
279 | if step_number == 1:
280 | # Initial pre-commit investigation tasks
281 | return [
282 | "Search for all git repositories in the specified path using appropriate tools",
283 | "Check git status to identify staged, unstaged, and untracked changes as required",
284 | "Execute git status to see all changes",
285 | "Execute git diff --cached for staged changes (exclude binary files)",
286 | "Execute git diff for unstaged changes (exclude binary files)",
287 | "List any relevant untracked files as well.",
288 | "Understand what functionality was added, modified, or removed",
289 | "Identify the scope and intent of the changes being committed",
290 | "CRITICAL: You are on step 1 - you MUST set next_step_required=True and continue to at least step 3 minimum",
291 | ]
292 | elif step_number == 2:
293 | # Need deeper investigation
294 | actions = [
295 | "Examine the specific files you've identified as changed or relevant",
296 | "Analyze the logic and implementation details of modifications",
297 | "Check for potential issues: bugs, security risks, performance problems",
298 | "Verify that changes align with good coding practices and patterns",
299 | "Look for missing tests, documentation, or configuration updates",
300 | ]
301 |
302 | # Add step validation reminder
303 | if request and request.total_steps >= 3:
304 | actions.append(
305 | f"CRITICAL: You are on step 2 of {request.total_steps} minimum steps - you MUST set next_step_required=True unless this is the final step"
306 | )
307 |
308 | return actions
309 | elif step_number >= 2 and (findings_count > 2 or issues_count > 0):
310 | # Close to completion - need final verification
311 | actions = [
312 | "Verify all identified issues have been properly documented",
313 | "Check for any missed dependencies or related files that need review",
314 | "Confirm the completeness and correctness of your assessment",
315 | "Ensure all security, performance, and quality concerns are captured",
316 | "Validate that your findings are comprehensive and actionable",
317 | ]
318 |
319 | # Add step validation reminder
320 | if request and request.total_steps >= 3 and step_number < request.total_steps:
321 | actions.append(
322 | f"CRITICAL: You are on step {step_number} of {request.total_steps} minimum steps - set next_step_required=True to continue"
323 | )
324 | elif request and request.total_steps >= 3 and step_number >= request.total_steps:
325 | actions.append(
326 | f"You are on final step {step_number} - you may now set next_step_required=False to complete"
327 | )
328 |
329 | return actions
330 | else:
331 | # General investigation needed
332 | actions = [
333 | "Continue examining the changes and their potential impact",
334 | "Gather more evidence using appropriate investigation tools",
335 | "Test your assumptions about the changes and their effects",
336 | "Look for patterns that confirm or refute your current assessment",
337 | ]
338 |
339 | # Add step validation reminder for all other cases
340 | if request and request.total_steps >= 3:
341 | if step_number < request.total_steps:
342 | actions.append(
343 | f"CRITICAL: You are on step {step_number} of {request.total_steps} minimum steps - set next_step_required=True to continue"
344 | )
345 | else:
346 | actions.append(
347 | f"You are on final step {step_number} - you may now set next_step_required=False to complete"
348 | )
349 |
350 | return actions
351 |
352 | def should_call_expert_analysis(self, consolidated_findings, request=None) -> bool:
353 | """
354 | Decide when to call external model based on investigation completeness.
355 |
356 | For continuations with external type, always proceed with expert analysis.
357 | """
358 | # Check if user requested to skip assistant model
359 | if request and not self.get_request_use_assistant_model(request):
360 | return False
361 |
362 | # For continuations with external type, always proceed with expert analysis
363 | continuation_id = self.get_request_continuation_id(request)
364 | if continuation_id and request.precommit_type == "external":
365 | return True # Always perform expert analysis for external continuations
366 |
367 | # Check if we have meaningful investigation data
368 | return (
369 | len(consolidated_findings.relevant_files) > 0
370 | or len(consolidated_findings.findings) >= 2
371 | or len(consolidated_findings.issues_found) > 0
372 | )
373 |
374 | def prepare_expert_analysis_context(self, consolidated_findings) -> str:
375 | """Prepare context for external model call for final pre-commit validation."""
376 | context_parts = [
377 | f"=== PRE-COMMIT ANALYSIS REQUEST ===\\n{self.initial_request or 'Pre-commit validation initiated'}\\n=== END REQUEST ==="
378 | ]
379 |
380 | # Add investigation summary
381 | investigation_summary = self._build_precommit_summary(consolidated_findings)
382 | context_parts.append(
383 | f"\\n=== AGENT'S PRE-COMMIT INVESTIGATION ===\\n{investigation_summary}\\n=== END INVESTIGATION ==="
384 | )
385 |
386 | # Add git configuration context if available
387 | if self.git_config:
388 | config_text = "\\n".join(f"- {key}: {value}" for key, value in self.git_config.items())
389 | context_parts.append(f"\\n=== GIT CONFIGURATION ===\\n{config_text}\\n=== END CONFIGURATION ===")
390 |
391 | # Add relevant methods/functions if available
392 | if consolidated_findings.relevant_context:
393 | methods_text = "\\n".join(f"- {method}" for method in consolidated_findings.relevant_context)
394 | context_parts.append(f"\\n=== RELEVANT CODE ELEMENTS ===\\n{methods_text}\\n=== END CODE ELEMENTS ===")
395 |
396 | # Add issues found evolution if available
397 | if consolidated_findings.issues_found:
398 | issues_text = "\\n".join(
399 | f"[{issue.get('severity', 'unknown').upper()}] {issue.get('description', 'No description')}"
400 | for issue in consolidated_findings.issues_found
401 | )
402 | context_parts.append(f"\\n=== ISSUES IDENTIFIED ===\\n{issues_text}\\n=== END ISSUES ===")
403 |
404 | # Add assessment evolution if available
405 | if consolidated_findings.hypotheses:
406 | assessments_text = "\\n".join(
407 | f"Step {h['step']}: {h['hypothesis']}" for h in consolidated_findings.hypotheses
408 | )
409 | context_parts.append(f"\\n=== ASSESSMENT EVOLUTION ===\\n{assessments_text}\\n=== END ASSESSMENTS ===")
410 |
411 | # Add images if available
412 | if consolidated_findings.images:
413 | images_text = "\\n".join(f"- {img}" for img in consolidated_findings.images)
414 | context_parts.append(
415 | f"\\n=== VISUAL VALIDATION INFORMATION ===\\n{images_text}\\n=== END VISUAL INFORMATION ==="
416 | )
417 |
418 | return "\\n".join(context_parts)
419 |
420 | def _build_precommit_summary(self, consolidated_findings) -> str:
421 | """Prepare a comprehensive summary of the pre-commit investigation."""
422 | summary_parts = [
423 | "=== SYSTEMATIC PRE-COMMIT INVESTIGATION SUMMARY ===",
424 | f"Total steps: {len(consolidated_findings.findings)}",
425 | f"Files examined: {len(consolidated_findings.files_checked)}",
426 | f"Relevant files identified: {len(consolidated_findings.relevant_files)}",
427 | f"Code elements analyzed: {len(consolidated_findings.relevant_context)}",
428 | f"Issues identified: {len(consolidated_findings.issues_found)}",
429 | "",
430 | "=== INVESTIGATION PROGRESSION ===",
431 | ]
432 |
433 | for finding in consolidated_findings.findings:
434 | summary_parts.append(finding)
435 |
436 | return "\\n".join(summary_parts)
437 |
438 | def should_include_files_in_expert_prompt(self) -> bool:
439 | """Include files in expert analysis for comprehensive validation."""
440 | return True
441 |
442 | def should_embed_system_prompt(self) -> bool:
443 | """Embed system prompt in expert analysis for proper context."""
444 | return True
445 |
446 | def get_expert_thinking_mode(self) -> str:
447 | """Use high thinking mode for thorough pre-commit analysis."""
448 | return "high"
449 |
450 | def get_expert_analysis_instruction(self) -> str:
451 | """Get specific instruction for pre-commit expert analysis."""
452 | return (
453 | "Please provide comprehensive pre-commit validation based on the investigation findings. "
454 | "Focus on identifying any remaining issues, validating the completeness of the analysis, "
455 | "and providing final recommendations for commit readiness."
456 | )
457 |
458 | # Hook method overrides for precommit-specific behavior
459 |
460 | def prepare_step_data(self, request) -> dict:
461 | """
462 | Map precommit-specific fields for internal processing.
463 | """
464 | step_data = {
465 | "step": request.step,
466 | "step_number": request.step_number,
467 | "findings": request.findings,
468 | "files_checked": request.files_checked,
469 | "relevant_files": request.relevant_files,
470 | "relevant_context": request.relevant_context,
471 | "issues_found": request.issues_found,
472 | "precommit_type": request.precommit_type,
473 | "hypothesis": request.findings, # Map findings to hypothesis for compatibility
474 | "images": request.images or [],
475 | "confidence": "high", # Dummy value for workflow_mixin compatibility
476 | }
477 | return step_data
478 |
479 | def should_skip_expert_analysis(self, request, consolidated_findings) -> bool:
480 | """
481 | Precommit workflow skips expert analysis only when precommit_type is "internal".
482 | Default is always to use expert analysis (external).
483 | For continuations with external type, always perform expert analysis immediately.
484 | """
485 | # If it's a continuation and precommit_type is external, don't skip
486 | continuation_id = self.get_request_continuation_id(request)
487 | if continuation_id and request.precommit_type != "internal":
488 | return False # Always do expert analysis for external continuations
489 |
490 | return request.precommit_type == "internal" and not request.next_step_required
491 |
492 | def store_initial_issue(self, step_description: str):
493 | """Store initial request for expert analysis."""
494 | self.initial_request = step_description
495 |
496 | # Override inheritance hooks for precommit-specific behavior
497 |
498 | def get_completion_status(self) -> str:
499 | """Precommit tools use precommit-specific status."""
500 | return "validation_complete_ready_for_commit"
501 |
502 | def get_completion_data_key(self) -> str:
503 | """Precommit uses 'complete_validation' key."""
504 | return "complete_validation"
505 |
506 | def get_final_analysis_from_request(self, request):
507 | """Precommit tools use 'findings' field."""
508 | return request.findings
509 |
510 | def get_precommit_type(self, request) -> str:
511 | """Get precommit type from request. Hook method for clean inheritance."""
512 | try:
513 | return request.precommit_type or "external"
514 | except AttributeError:
515 | return "external" # Default to external validation
516 |
517 | def get_consolidated_issues_count(self) -> int:
518 | """Get count of issues from consolidated findings. Hook method for clean access."""
519 | try:
520 | return len(self.consolidated_findings.issues_found)
521 | except AttributeError:
522 | return 0
523 |
524 | def get_completion_message(self) -> str:
525 | """Precommit-specific completion message."""
526 | return (
527 | "Pre-commit validation complete. You have identified all issues "
528 | "and verified commit readiness. MANDATORY: Present the user with the complete validation results "
529 | "and IMMEDIATELY proceed with commit if no critical issues found, or provide specific fix guidance "
530 | "if issues need resolution. Focus on actionable next steps."
531 | )
532 |
533 | def get_skip_reason(self) -> str:
534 | """Precommit-specific skip reason."""
535 | return (
536 | "Completed comprehensive pre-commit validation with internal analysis only (no external model validation)"
537 | )
538 |
539 | def get_skip_expert_analysis_status(self) -> str:
540 | """Precommit-specific expert analysis skip status."""
541 | return "skipped_due_to_internal_analysis_type"
542 |
543 | def prepare_work_summary(self) -> str:
544 | """Precommit-specific work summary."""
545 | return self._build_precommit_summary(self.consolidated_findings)
546 |
547 | def get_completion_next_steps_message(self, expert_analysis_used: bool = False) -> str:
548 | """
549 | Precommit-specific completion message.
550 |
551 | Args:
552 | expert_analysis_used: True if expert analysis was successfully executed
553 | """
554 | base_message = (
555 | "PRE-COMMIT VALIDATION IS COMPLETE. You may delete any `zen_precommit.changeset` created. You MUST now summarize "
556 | "and present ALL validation results, identified issues with their severity levels, and exact commit recommendations. "
557 | "Clearly state whether the changes are ready for commit or require fixes first. Provide concrete, actionable guidance for "
558 | "any issues that need resolution—make it easy for a developer to understand exactly what needs to be "
559 | "done before committing."
560 | )
561 |
562 | # Add expert analysis guidance only when expert analysis was actually used
563 | if expert_analysis_used:
564 | expert_guidance = self.get_expert_analysis_guidance()
565 | if expert_guidance:
566 | return f"{base_message}\n\n{expert_guidance}"
567 |
568 | return base_message
569 |
570 | def get_expert_analysis_guidance(self) -> str:
571 | """
572 | Get additional guidance for handling expert analysis results in pre-commit context.
573 |
574 | Returns:
575 | Additional guidance text for validating and using expert analysis findings
576 | """
577 | return (
578 | "IMPORTANT: Expert analysis has been provided above. You MUST carefully review "
579 | "the expert's validation findings and security assessments. Cross-reference the "
580 | "expert's analysis with your own investigation to ensure all critical issues are "
581 | "addressed. Pay special attention to any security vulnerabilities, performance "
582 | "concerns, or architectural issues identified by the expert review."
583 | )
584 |
585 | def get_step_guidance_message(self, request) -> str:
586 | """
587 | Precommit-specific step guidance with detailed investigation instructions.
588 | """
589 | step_guidance = self.get_precommit_step_guidance(request.step_number, request)
590 | return step_guidance["next_steps"]
591 |
592 | def get_precommit_step_guidance(self, step_number: int, request) -> dict[str, Any]:
593 | """
594 | Provide step-specific guidance for precommit workflow.
595 | Uses get_required_actions to determine what needs to be done,
596 | then formats those actions into appropriate guidance messages.
597 | """
598 | # Get the required actions from the single source of truth
599 | required_actions = self.get_required_actions(
600 | step_number,
601 | request.precommit_type or "external", # Using precommit_type as confidence proxy
602 | request.findings or "",
603 | request.total_steps,
604 | request, # Pass request for continuation-aware decisions
605 | )
606 |
607 | # Check if this is a continuation to provide context-aware guidance
608 | continuation_id = self.get_request_continuation_id(request)
609 | is_external_continuation = continuation_id and request.precommit_type == "external"
610 | is_internal_continuation = continuation_id and request.precommit_type == "internal"
611 |
612 | # Format the guidance based on step number and continuation status
613 | if step_number == 1:
614 | if is_external_continuation:
615 | # Fast-track mode for external continuations
616 | next_steps = (
617 | "You are on step 1 of MAXIMUM 2 steps. CRITICAL: Gather and save the complete git changeset NOW. "
618 | "MANDATORY ACTIONS:\\n"
619 | + "\\n".join(f"{i+1}. {action}" for i, action in enumerate(required_actions))
620 | + "\\n\\nMANDATORY: The changeset may be large. You MUST save the required changeset as a 'zen_precommit.changeset' file "
621 | "(replacing any existing one) in your work directory and include the FULL absolute path in relevant_files (exclude any "
622 | "binary files). ONLY include the code changes, no extra commentary."
623 | "Set next_step_required=True and step_number=2 for the next call."
624 | )
625 | elif is_internal_continuation:
626 | # Internal validation mode
627 | next_steps = (
628 | "Continuing previous conversation with internal validation only. The analysis will build "
629 | "upon the prior findings without external model validation. REQUIRED ACTIONS:\\n"
630 | + "\\n".join(f"{i+1}. {action}" for i, action in enumerate(required_actions))
631 | )
632 | else:
633 | # Normal flow for new validations
634 | next_steps = (
635 | f"MANDATORY: DO NOT call the {self.get_name()} tool again immediately. You MUST first investigate "
636 | f"the git repositories and changes using appropriate tools. CRITICAL AWARENESS: You need to:\\n"
637 | + "\\n".join(f"{i+1}. {action}" for i, action in enumerate(required_actions))
638 | + f"\\n\\nOnly call {self.get_name()} again AFTER completing your investigation. "
639 | f"When you call {self.get_name()} next time, use step_number: {step_number + 1} "
640 | f"and report specific files examined, changes analyzed, and validation findings discovered."
641 | )
642 |
643 | elif step_number == 2:
644 | # CRITICAL: Check if violating minimum step requirement
645 | if (
646 | request.total_steps >= 3
647 | and request.step_number < request.total_steps
648 | and not request.next_step_required
649 | ):
650 | next_steps = (
651 | f"ERROR: You set total_steps={request.total_steps} but next_step_required=False on step {request.step_number}. "
652 | f"This violates the minimum step requirement. You MUST set next_step_required=True until you reach the final step. "
653 | f"Call {self.get_name()} again with next_step_required=True and continue your investigation."
654 | )
655 | elif is_external_continuation or (not request.next_step_required and request.precommit_type == "external"):
656 | # Fast-track completion or about to complete - ensure changeset is saved
657 | next_steps = (
658 | "Proceeding immediately to expert analysis. "
659 | f"MANDATORY: call {self.get_name()} tool immediately again, and set next_step_required=False to "
660 | f"trigger external validation NOW. "
661 | f"MANDATORY: Include the entire changeset! The changeset may be large. You MUST save the required "
662 | f"changeset as a 'zen_precommit.changeset' file (replacing any existing one) in your work directory "
663 | f"and include the FULL absolute path in relevant_files so the expert can access the complete changeset. "
664 | f"ONLY include the code changes, no extra commentary."
665 | )
666 | else:
667 | # Normal flow - deeper analysis needed
668 | next_steps = (
669 | f"STOP! Do NOT call {self.get_name()} again yet. You are on step 2 of {request.total_steps} minimum required steps. "
670 | f"MANDATORY ACTIONS before calling {self.get_name()} step {step_number + 1}:\\n"
671 | + "\\n".join(f"{i+1}. {action}" for i, action in enumerate(required_actions))
672 | + f"\\n\\nRemember: You MUST set next_step_required=True until step {request.total_steps}. "
673 | + f"Only call {self.get_name()} again with step_number: {step_number + 1} AFTER completing these validations."
674 | )
675 |
676 | elif step_number >= 3:
677 | if not request.next_step_required and request.precommit_type == "external":
678 | # About to complete - ensure changeset is saved
679 | next_steps = (
680 | "Completing validation and proceeding to expert analysis. "
681 | "MANDATORY: Save the complete git changeset as a 'zen_precommit.changeset' file "
682 | "in your work directory and include the FULL absolute path in relevant_files."
683 | )
684 | else:
685 | # Later steps - final verification
686 | next_steps = (
687 | f"WAIT! Your validation needs final verification. DO NOT call {self.get_name()} immediately. REQUIRED ACTIONS:\\n"
688 | + "\\n".join(f"{i+1}. {action}" for i, action in enumerate(required_actions))
689 | + f"\\n\\nREMEMBER: Ensure you have identified all potential issues and verified commit readiness. "
690 | f"Document findings with specific file references and issue descriptions, then call {self.get_name()} "
691 | f"with step_number: {step_number + 1}."
692 | )
693 | else:
694 | # Fallback for any other case - check minimum step violation first
695 | if (
696 | request.total_steps >= 3
697 | and request.step_number < request.total_steps
698 | and not request.next_step_required
699 | ):
700 | next_steps = (
701 | f"ERROR: You set total_steps={request.total_steps} but next_step_required=False on step {request.step_number}. "
702 | f"This violates the minimum step requirement. You MUST set next_step_required=True until step {request.total_steps}."
703 | )
704 | elif not request.next_step_required and request.precommit_type == "external":
705 | next_steps = (
706 | "Completing validation. "
707 | "MANDATORY: Save complete git changeset as 'zen_precommit.changeset' file and include path in relevant_files, "
708 | "excluding any binary files."
709 | )
710 | else:
711 | next_steps = (
712 | f"PAUSE VALIDATION. Before calling {self.get_name()} step {step_number + 1}, you MUST examine more code and changes. "
713 | + "Required: "
714 | + ", ".join(required_actions[:2])
715 | + ". "
716 | + f"Your next {self.get_name()} call (step_number: {step_number + 1}) must include "
717 | f"NEW evidence from actual change analysis, not just theories. NO recursive {self.get_name()} calls "
718 | f"without investigation work!"
719 | )
720 |
721 | return {"next_steps": next_steps}
722 |
723 | def customize_workflow_response(self, response_data: dict, request) -> dict:
724 | """
725 | Customize response to match precommit workflow format.
726 | """
727 | # Store initial request on first step
728 | if request.step_number == 1:
729 | self.initial_request = request.step
730 | # Store git configuration for expert analysis
731 | if request.path:
732 | self.git_config = {
733 | "path": request.path,
734 | "compare_to": request.compare_to,
735 | "include_staged": request.include_staged,
736 | "include_unstaged": request.include_unstaged,
737 | "severity_filter": request.severity_filter,
738 | }
739 |
740 | # Convert generic status names to precommit-specific ones
741 | tool_name = self.get_name()
742 | status_mapping = {
743 | f"{tool_name}_in_progress": "validation_in_progress",
744 | f"pause_for_{tool_name}": "pause_for_validation",
745 | f"{tool_name}_required": "validation_required",
746 | f"{tool_name}_complete": "validation_complete",
747 | }
748 |
749 | if response_data["status"] in status_mapping:
750 | response_data["status"] = status_mapping[response_data["status"]]
751 |
752 | # Rename status field to match precommit workflow
753 | if f"{tool_name}_status" in response_data:
754 | response_data["validation_status"] = response_data.pop(f"{tool_name}_status")
755 | # Add precommit-specific status fields
756 | response_data["validation_status"]["issues_identified"] = len(self.consolidated_findings.issues_found)
757 | response_data["validation_status"]["precommit_type"] = request.precommit_type or "external"
758 |
759 | # Map complete_precommitworkflow to complete_validation
760 | if f"complete_{tool_name}" in response_data:
761 | response_data["complete_validation"] = response_data.pop(f"complete_{tool_name}")
762 |
763 | # Map the completion flag to match precommit workflow
764 | if f"{tool_name}_complete" in response_data:
765 | response_data["validation_complete"] = response_data.pop(f"{tool_name}_complete")
766 |
767 | return response_data
768 |
769 | # Required abstract methods from BaseTool
770 | def get_request_model(self):
771 | """Return the precommit workflow-specific request model."""
772 | return PrecommitRequest
773 |
774 | async def prepare_prompt(self, request) -> str:
775 | """Not used - workflow tools use execute_workflow()."""
776 | return "" # Workflow tools use execute_workflow() directly
777 |
```