3.0 KiB
3.0 KiB
Session Instructions for Claude
This file contains mandatory instructions for EVERY conversation session.
🎯 Mandatory Tooling Usage Policy
CRITICAL: Claude Code must maximize the use of available advanced features for efficiency and quality.
At the START of EVERY Task:
Provide a Tooling Strategy Decision:
### 🎯 Tooling Strategy Decision
**Task Analysis**: [Brief description of the task]
**Tooling Decisions**:
- **Agents**: Using [agent-name] / Not using - Reason: [specific justification]
- **Slash Commands**: Using [/command] / Not using - Reason: [specific justification]
- **MCP Servers**: Using [server: tool] / Not using - Reason: [specific justification]
- **Approach**: [Overall strategy for completing the task]
At the END of EVERY Task:
Provide a Task Completion Summary:
### 📊 Task Completion Summary
**What Was Done**: [Brief description]
**Features Involved**:
- Agents: [List or None with justification]
- Slash Commands: [List or None with justification]
- MCP Servers: [List or None with justification]
- Core Tools: [List]
- Files Modified: [List]
- Performance: [Notes]
**Efficiency Notes**: [Observations]
📋 Available Resources
Agents (8 total)
- Explore: Codebase exploration (quick/medium/thorough)
- Plan: Planning and design
- test-engineer: Generate comprehensive tests
- code-reviewer: Code quality reviews
- refactoring-specialist: Code cleanup
- debugger: Bug diagnosis
- architect: System design
- documentation-writer: Comprehensive docs
- security-analyst: Security reviews
MCP Servers (8 total)
- serena: Code navigation, symbol search, memory
- sequential-thinking: Complex reasoning
- context7: Library documentation
- memory: Knowledge graph
- fetch: Web content retrieval
- windows-mcp: Desktop automation
- playwright: Browser automation
- database-server: SQL access
Slash Commands (9 total)
/test [file]: Generate and run tests/review [file]: Code review/explain [file]: Explain code/analyze [path]: Code analysis/optimize [file]: Performance optimization/implement [desc]: Feature implementation/scaffold [type]: Generate boilerplate/adr [action]: Manage ADRs/setup-info: Display setup info
⚠️ When NOT to Use Advanced Features
Only skip agents/slash commands/MCP when:
- Single file reads with known path
- Simple edits to existing code
- Tasks completable in 1-2 tool calls
- Purely conversational/informational requests
Always state explicitly if skipping: "Not using [feature] because [reason]"
🎯 Project Context
- Project: Foundry VTT v11.315 + PF1e System v10.8
- Purpose: Macro development and game system debugging
- Main Files: src/macro.js, src/macro_haste.js, CLAUDE.md
- Documentation: See CLAUDE.md for full project details
Reference: See CLAUDE.md for complete project documentation