mirror of
https://github.com/catlog22/Claude-Code-Workflow.git
synced 2026-02-05 01:50:27 +08:00
- Introduced a new test script for the CodexLens LSP tool to validate core functionalities including symbol search, find definition, find references, and get hover. - Created comprehensive documentation for the MCP endpoint design, detailing the architecture, features, and integration with the CCW MCP Manager. - Developed a detailed implementation plan for transitioning to a real LSP server, outlining phases, architecture, and acceptance criteria.
3.1 KiB
3.1 KiB
Claude Instructions
- Coding Philosophy: @~/.claude/workflows/coding-philosophy.md
CLI Endpoints
- CLI Tools Usage: @~/.claude/workflows/cli-tools-usage.md
- CLI Endpoints Config: @~/.claude/cli-tools.json
Strictly follow the cli-tools.json configuration
Available CLI endpoints are dynamically defined by the config file:
- Built-in tools and their enable/disable status
- Custom API endpoints registered via the Dashboard
- Managed through the CCW Dashboard Status page
Tool Execution
- Context Requirements: @~/.claude/workflows/context-tools.md
- File Modification: @~/.claude/workflows/file-modification.md
Agent Calls
- Always use
run_in_background: falsefor Task tool agent calls:Task({ subagent_type: "xxx", prompt: "...", run_in_background: false })to ensure synchronous execution and immediate result visibility - TaskOutput usage: Only use
TaskOutput({ task_id: "xxx", block: false })+ sleep loop to poll completion status. NEVER read intermediate output during agent/CLI execution - wait for final result only
CLI Tool Calls (ccw cli)
- Default:
run_in_background: true- Unless otherwise specified, always use background execution for CLI calls:Bash({ command: "ccw cli -p '...' --tool gemini", run_in_background: true }) - After CLI call: Stop output immediately - let CLI execute in background. DO NOT use TaskOutput polling - wait for hook callback to receive results
CLI Analysis Calls
- Wait for results: MUST wait for CLI analysis to complete before taking any write action. Do NOT proceed with fixes while analysis is running
- Value every call: Each CLI invocation is valuable and costly. NEVER waste analysis results:
- Aggregate multiple analysis results before proposing solutions
CLI Auto-Invoke Triggers
Proactive CLI invocation scenarios - Auto-invoke ccw cli for external analysis in these cases:
| Trigger Condition | Recommended Mode | Description |
|---|---|---|
| Bug fix fails after 1+ attempts | --mode analysis --rule analysis-diagnose-bug-root-cause |
Invoke CLI for root cause analysis when self-repair attempts fail |
| Unclear task description | --mode analysis --rule planning-breakdown-task-steps |
Invoke CLI for task decomposition when requirements are ambiguous |
| Quick planning needed | --mode analysis --rule planning-plan-architecture-design |
Invoke CLI for architecture design on complex feature requests |
| Uncertain code patterns | --mode analysis --rule analysis-analyze-code-patterns |
Invoke CLI to analyze existing code style/patterns when uncertain |
| Security/performance critical paths | --mode analysis --rule analysis-assess-security-risks |
Proactively request review for security or performance-sensitive code |
Execution principles:
- When trigger conditions are met, invoke CLI without user confirmation
- Wait for results before determining next steps
- Tool fallback chain:
gemini→qwen→codex
Code Diagnostics
- Prefer
mcp__ide__getDiagnosticsfor code error checking over shell-based TypeScript compilation