Skip to content

claude-code-docs-map.md

Latest commit

 

History

History
359 lines (278 loc) · 15 KB

claude-code-docs-map.md

File metadata and controls

359 lines (278 loc) · 15 KB
title Claude Code Documentation Map
category tools
type documentation
status active
created 2025-10-25
tags
claude-code
documentation
reference
self-discovery
related
claude-trace
sentinel-mcp-capabilities
sub-agents
code_files
N/A (external documentation)
source https://docs.claude.com/en/docs/claude-code/claude_code_docs_map.md

Claude Code Documentation Map

Purpose: Markdown index of all official Claude Code documentation. Enables Claude Code to self-reference capabilities via WebFetch tool.

Pattern: Follows llms.txt convention for AI-readable documentation indexes.

How It Works

Auto-Discovery Flow:

User asks: "Can Claude Code do X?"
    ↓
System prompt triggers WebFetch
    ↓
Fetches: claude_code_docs_map.md
    ↓
Parses map for relevant docs
    ↓
Fetches specific doc pages
    ↓
Answers based on current docs

Key Benefit: Overcomes LLM knowledge cutoff - Claude Code always has access to latest documentation, even for features added after training data cutoff.

Complete Documentation Index

Getting Started

Document URL Purpose
Overview /overview.md What Claude Code does, why developers use it
Quickstart /quickstart.md Installation through first workflows
Common Workflows /common-workflows.md Understand codebases, fix bugs, refactor
Claude Code on Web /claude-code-on-the-web.md Web usage, cloud environments, network security

Build with Claude Code

Document URL Purpose
Sub-agents /sub-agents.md Configure specialized agent delegation
Plugins /plugins.md Create and manage plugin extensions
Skills /skills.md Build agent skills with restricted tools
Output Styles /output-styles.md Customize response formatting
Hooks Guide /hooks-guide.md Event-triggered automation
Headless /headless.md API usage without interactive terminal
GitHub Actions /github-actions.md CI/CD integration for GitHub
GitLab CI/CD /gitlab-ci-cd.md Pipeline integration for GitLab
MCP /mcp.md Model Context Protocol servers

Claude Agent SDK

Document URL Purpose
Migration Guide /sdk/migration-guide.md Upgrade SDK with breaking changes

Deployment

Document URL Purpose
Third-Party Integrations /third-party-integrations.md Cloud providers, infrastructure
Amazon Bedrock /amazon-bedrock.md AWS-hosted deployment
Google Vertex AI /google-vertex-ai.md GCP-hosted deployment
Network Config /network-config.md Proxy and certificate setup
LLM Gateway /llm-gateway.md LiteLLM routing setup
Dev Container /devcontainer.md Containerized development
Sandboxing /sandboxing.md Security isolation features

Administration

Document URL Purpose
Setup /setup.md Installation, system requirements
IAM /iam.md Authentication, access control
Security /security.md Injection protection, best practices
Data Usage /data-usage.md Training policy, retention
Monitoring Usage /monitoring-usage.md Metrics, event tracking
Costs /costs.md Token usage, reduction strategies
Analytics /analytics.md Team insights, adoption metrics
Plugin Marketplaces /plugin-marketplaces.md Plugin distribution

Configuration

Document URL Purpose
Settings /settings.md Config files, env variables
VS Code /vs-code.md VSCode extension (Beta)
JetBrains /jetbrains.md IntelliJ-based IDE integration
Terminal Config /terminal-config.md Appearance, notifications
Model Config /model-config.md Model selection, behavior
Memory /memory.md CLAUDE.md project memories
Status Line /statusline.md Custom status display

Reference

Document URL Purpose
CLI Reference /cli-reference.md Commands, flags
Interactive Mode /interactive-mode.md Keyboard shortcuts, vim mode
Slash Commands /slash-commands.md Built-in and custom commands
Checkpointing /checkpointing.md Session state tracking
Hooks /hooks.md Hook event specification
Plugins Reference /plugins-reference.md Plugin manifest schema
Troubleshooting /troubleshooting.md Installation, performance, IDE issues

Resources

Document URL Purpose
Legal and Compliance /legal-and-compliance.md Licensing, BAA, security reporting

Priority Documentation for Soulfield Integration

Essential Reads

1. Sub-agents (/sub-agents.md)

  • Why: Soulfield uses sub-agent pattern extensively (CLI delegation, token extension)
  • Key Topics: Configuration, delegation patterns, "don't report back" rule
  • Soulfield Context: Unlimited token capacity via sub-agent spawning

2. MCP (/mcp.md)

  • Why: 8 MCP servers integrated (Supabase, Google Workspace, Playwright, etc.)
  • Key Topics: Server configuration, tool invocation, security
  • Soulfield Context: Read [[sentinel-mcp-capabilities]] for full list

3. Memory (/memory.md)

  • Why: CLAUDE.md and .claude/CLAUDE.md both in use
  • Key Topics: Project memory, global memory, memory hierarchies
  • Soulfield Context: /home/michael/.claude/CLAUDE.md = global, CLAUDE.md = project

4. Hooks Guide (/hooks-guide.md)

  • Why: Potential for automating lens validation, git commits
  • Key Topics: Event types, hook scripts, examples
  • Soulfield Context: Could trigger lens-validator on pre-commit

5. Output Styles (/output-styles.md)

  • Why: Sentinel output style enforces lens contract
  • Key Topics: Custom formatting, structured responses
  • Soulfield Context: /home/michael/.claude/output-styles/sentinel.md

Advanced Topics

6. Headless (/headless.md)

  • Use Case: Integrate Claude Code into Soulfield agent workflows
  • Example: Agent triggers Claude Code session programmatically

7. GitHub Actions (/github-actions.md)

  • Use Case: CI/CD pipeline for lens validation, test running
  • Example: Auto-run lens-enforcer on PR creation

8. Sandboxing (/sandboxing.md)

  • Use Case: Isolate dangerous operations (file writes, bash commands)
  • Example: Run agents in sandboxed environment for security

How to Use This Map

Pattern 1: Direct Reference (During Conversation)

User: "Can Claude Code integrate with GitHub Actions?"
Sentinel: *auto-fetches /github-actions.md*
Sentinel: "Yes - see github-actions.md for CI/CD integration guide..."

Pattern 2: Manual Lookup (Development)

# Fetch specific doc
curl https://docs.claude.com/en/docs/claude-code/sub-agents.md

# Or use WebFetch tool in Claude Code
claude
> Fetch docs for sub-agents

Pattern 3: Batch Download (Offline Reference)

# Download all docs to local directory
mkdir -p ~/soulfield/docs/claude-code
cd ~/soulfield/docs/claude-code

# Fetch docs map
curl -O https://docs.claude.com/en/docs/claude-code/claude_code_docs_map.md

# Fetch all referenced docs (requires parsing map)
# TODO: Create script to batch download

Integration with Soulfield Memory Systems

Documentation Storage Strategy

3-Tier Documentation:

  1. Official Docs (Claude Code) → Fetch on-demand via WebFetch
  2. Soulfield Reference (Obsidian) → workspace/docs/Obsidian/Knowledge/reference/
  3. Agent Training Data (Supabase) → backend/services/memory/memory-supabase.cjs

Workflow:

User asks about Claude Code feature
    ↓
Sentinel fetches official docs (WebFetch)
    ↓
Creates/updates Soulfield reference file (Obsidian)
    ↓
Optionally: Embed in Supabase for agent recall

Example:

# User asks about sub-agents
claude
> How do sub-agents work?

# Sentinel:
# 1. Fetches /sub-agents.md via WebFetch
# 2. Reads existing workspace/docs/Obsidian/plans/integrations/claude-code/...
# 3. Creates summary in Obsidian
# 4. Answers with citations to official docs

Gaps in Official Documentation

Not Covered in Docs Map:

  • ❌ Token usage optimization strategies (only "costs" doc)
  • ❌ Sub-agent performance benchmarks
  • ❌ MCP server development guide (only usage)
  • ❌ Advanced lens framework patterns
  • ❌ Multi-agent orchestration examples

Soulfield Extensions (Fill Gaps):

  • ✅ [[sentinel-agent-systems]] - 3 agent systems explained
  • ✅ [[sentinel-mcp-capabilities]] - All 8 MCP servers documented
  • ✅ [[claude-trace]] - Debugging tool not in official docs
  • ✅ [[lens-contract]] - Truth-first engineering patterns

Monitoring Documentation Updates

Official Docs Change Detection:

# Check docs map for updates
curl -s https://docs.claude.com/en/docs/claude-code/claude_code_docs_map.md | md5sum

# Compare with last known version
echo "a1b2c3d4..." > /tmp/last-docs-hash
curl -s https://docs.claude.com/en/docs/claude-code/claude_code_docs_map.md | md5sum > /tmp/current-docs-hash

# Alert on change
diff /tmp/last-docs-hash /tmp/current-docs-hash && echo "Docs updated!"

Automation Opportunity:

  • Create cron job to check for docs updates
  • Auto-update Soulfield reference files when changes detected
  • Log updates to Obsidian daily notes

Quick Reference: Most Used Docs

Task Document
Configure MCP servers /mcp.md
Use sub-agents /sub-agents.md
Set up hooks /hooks-guide.md
Customize output /output-styles.md
Manage memory /memory.md
Troubleshoot issues /troubleshooting.md
Debug with trace (not in official docs - see [[claude-trace]])

Example: Self-Discovery in Action

Real Conversation:

User: "Does Claude Code support GitLab CI/CD?"

Sentinel (internal):
1. Detects question about Claude Code feature
2. Triggers WebFetch to fetch docs map
3. Finds /gitlab-ci-cd.md in map
4. Fetches full GitLab CI/CD doc
5. Parses doc for answer

Sentinel (response):
"Yes - Claude Code supports GitLab CI/CD integration. See the official guide at:
https://docs.claude.com/en/docs/claude-code/gitlab-ci-cd.md

Key features:
- Pipeline integration
- Merge request automation
- Custom CI/CD variables
- Docker executor support

Would you like me to fetch the full guide?"

Without Docs Map:

User: "Does Claude Code support GitLab CI/CD?"

Sentinel (old behavior):
"[UNKNOWN] - My training data is from Jan 2025, GitLab CI/CD support may have been added since then. Check official docs at https://docs.claude.com/"

Impact: Accurate, current answers vs. admitting uncertainty

Related Documentation

Soulfield-Specific:

  • [[claude-trace]] - HTTP proxy for debugging Claude Code
  • [[sentinel-mcp-capabilities]] - All 8 MCP servers documented
  • [[sentinel-agent-systems]] - 3 agent systems explained
  • [[sub-agents-delegation-pattern]] - Unlimited token capacity via sub-agents

Official Claude Code:

Last Updated: 2025-10-25 Status: Active (official docs map maintained by Anthropic) Access: Public (no authentication required)