update-docs

You have just completed work on the current project. Analyze changes based on the provided context and automatically update relevant documentation.

Auto-Loaded Project Context:

@/CLAUDE.md @/docs/ai-context/project-structure.md @/docs/ai-context/docs-overview.md

Core Documentation Principle: Document Current State Only

CRITICAL: Always document the current "is" state of the system. Never reference legacy implementations, describe improvements made, or explain what changed. Documentation should read as if the current implementation has always existed.

Documentation Anti-Patterns to Avoid:

• ❌ "Refactored the voice pipeline to use streaming instead of batch processing"
• ❌ "Improved performance by implementing caching"
• ❌ "Previously used X, now uses Y for better results"
• ❌ "Legacy implementation has been replaced with..."

Documentation Best Practices:

• ✅ "The voice pipeline uses streaming for real-time processing"
• ✅ "Implements caching for frequently accessed data"
• ✅ "Uses Y for optimal results"
• ✅ "The system architecture follows..."

Step 1: Analyze Changes Based on Input

Determine Analysis Mode:

• No input (default): Analyze recent conversation context
• Git commit ID (e.g., "3b8d24e" or full hash): Analyze specific commit
• "uncommitted"/"staged"/"working": Analyze uncommitted changes
• "last N commits" (e.g., "last 3 commits"): Analyze recent commits

Execute Analysis:

Based on the input parameter:

For Git Commit Analysis:

# Get commit details
git show --name-status [COMMIT_ID]
git diff [COMMIT_ID]^ [COMMIT_ID]

For Uncommitted Changes:

# Get staged and unstaged changes
git status --porcelain
git diff HEAD
git diff --cached

For Recent Commits:

# Get recent commit history
git log --oneline -n [N]
git diff HEAD~[N] HEAD

For Session Context (default):

Review your recent conversation and tool usage for significant changes.

Look for Evidence of Documentation-Relevant Changes: