Agent skill
three-tier-docs
Use after code changes. Syncs CLAUDE.md, CONTEXT.md, and docs/ai-context/ automatically.
Install this agent skill to your Project
npx add-skill https://github.com/majiayu000/claude-skill-registry/tree/main/skills/data/three-tier-docs
SKILL.md
SKILL: Three-Tier Documentation
Purpose: Auto-sync documentation across 3 tiers (CLAUDE.md → CONTEXT.md → docs/) Target: Documenter agent after implementation
Quick Start
When to Use This Skill
- After implementing features
- After refactoring code
- After creating new components
- Run via
/documentcommand
Quick Reference
# Tier 1: Sync 3 Entry Point documents
# - CLAUDE.md (≤200 lines)
# - docs/ai-context/project-structure.md
# - docs/ai-context/docs-overview.md
# Tier 2: Component CONTEXT.md (≤200 lines)
# - Purpose, key files, patterns
# Tier 3: Feature CONTEXT.md (≤200 lines)
# - Implementation details
What This Skill Covers
In Scope
- 3-tier documentation hierarchy (CLAUDE.md, CONTEXT.md, docs/)
- Size limits and content organization
- Auto-sync patterns after code changes
Out of Scope
- Detailed templates → @.claude/skills/three-tier-docs/REFERENCE.md
- Verification skills → docs-verify
- Git documentation workflow → @.claude/skills/git-master/SKILL.md
Core Concepts
3-Tier Hierarchy
Tier 1: Entry Points (3 files only)
CLAUDE.md- Architecture, features, Quick Start (≤200 lines)docs/ai-context/project-structure.md- Tech stack, file tree (≤200 lines)docs/ai-context/docs-overview.md- Documentation navigation, Tier mapping (≤200 lines)- Required: CLAUDE.md top section must reference the other 2 files
- Constraint: docs/ai-context/ folder must contain exactly 2 files
Tier 2: CONTEXT.md (Component Directory)
- Purpose: Component context
- Content: Purpose, key files, patterns
- Size: ≤200 lines per file
- Scope: Component-level context
Tier 3: CONTEXT.md (Feature Directory)
- Purpose: Feature implementation details
- Content: Implementation details, deep-dive
- Size: ≤200 lines per file
- Scope: Feature-level context
Content Organization
| Tier | Location | Files | Update Frequency |
|---|---|---|---|
| 1 | Root + docs/ai-context/ | CLAUDE.md, project-structure.md, docs-overview.md | Project changes |
| 2 | Component dirs | CONTEXT.md | Component changes |
| 3 | Feature dirs | CONTEXT.md | Feature changes |
Execution Steps
Step 1: Sync Tier 1 Documents
Tier 1 Structure (3 files total):
CLAUDE.md- Project architecture, features, Quick Start (≤200 lines)docs/ai-context/project-structure.md- Tech stack, file treedocs/ai-context/docs-overview.md- Documentation navigation, Tier mapping
Required: CLAUDE.md must reference project-structure.md and docs-overview.md at the top
Verification:
# Check Tier 1 files exist
for file in "CLAUDE.md" "docs/ai-context/project-structure.md" "docs/ai-context/docs-overview.md"; do
if [ ! -f "$file" ]; then
echo "FAIL: Missing Tier 1 file: $file"
exit 1
fi
done
# Check CLAUDE.md references the other 2 files
if ! grep -q "project-structure.md" CLAUDE.md || ! grep -q "docs-overview.md" CLAUDE.md; then
echo "FAIL: CLAUDE.md must reference project-structure.md and docs-overview.md"
exit 1
fi
echo "✓ Tier 1 documents verified"
Step 2: Generate Component CONTEXT.md Files
Target directories: Any directory with code files (e.g., src/, components/, lib/)
Script: See REFERENCE.md for full bash implementation
# Generate CONTEXT.md for key directories
for dir in src/ components/ lib/ .claude/commands/ .claude/skills/ .claude/agents/; do
[ -d "$dir" ] || continue
[ -f "$dir/CONTEXT.md" ] && [ -s "$dir/CONTEXT.md" ] && continue
# Generate template (see REFERENCE.md for full script)
echo "✓ Generated $dir/CONTEXT.md"
done
Size Limit: ≤200 lines per file
Step 3: Verify Documentation Compliance
Invoke the docs-verify skill for comprehensive validation.
Validation includes:
- Tier 1 line limits (≤200 lines): CLAUDE.md, project-structure.md, docs-overview.md
- ai-context file count (exactly 2 files)
- Cross-reference validation
- Circular reference detection
Script: See @.claude/skills/docs-verify/SKILL.md for full verification commands
Step 4: Inline Verification
Verify documentation compliance immediately after update:
# 1. Tier 1 Line Limits (≤200)
for file in CLAUDE.md docs/ai-context/*.md; do
[ -f "$file" ] || continue
lines=$(wc -l < "$file" | tr -d ' ')
[ "$lines" -gt 200 ] && echo "FAIL: $file has $lines lines (limit: 200)"
done
# 2. ai-context file count (exactly 2)
count=$(find docs/ai-context -maxdepth 1 -name "*.md" -type f | wc -l | tr -d ' ')
[ "$count" -ne 2 ] && echo "FAIL: docs/ai-context/ has $count files (expected: 2)"
# 3. Cross-reference check
grep -oE '@[^][:space:]]+' CLAUDE.md | while read ref; do
[ ! -e "${ref#@}" ] && echo "Broken: $ref"
done
Verification Failure Recovery
Line count violations: CLAUDE.md >200 lines → Extract to docs/ai-context/ | CONTEXT.md >200 lines → Move examples or simplify
Broken cross-references: Check file exists with test -f {path} | Use absolute paths (e.g., @.claude/skills/...)
Missing frontmatter: Add required fields (name, description) | Validate with yamllint {file}
Recovery Steps: Invoke docs-verify skill → Find violating file → Apply fix from REFERENCE.md → Re-verify
Further Reading
Internal: @.claude/skills/three-tier-docs/REFERENCE.md - Complete templates, examples, verification patterns | @.claude/skills/documentation-best-practices/SKILL.md - Size limits, quality standards
External: Documentation System Design | Writing for AI
Version: claude-pilot 4.4.15
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
agent-ops-spec
Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.
agent-ops-state
Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.
agent-ops-spec
Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.
agent-ops-testing
Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.
agent-ops-testing
Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.
agent-ops-state
Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.
Didn't find tool you were looking for?