sync-claude（sync Claude 搜索）

Claude Documentation Sync

Validate all Claude-related documentation before commits. Default mode is changed (branch-scoped); use full for exhaustive validation.

Execution Flow

Phase 1: Discovery

Full mode: Find all Claude docs in repository:

• **/CLAUDE.md (excluding .terragrunt-cache/, .terragrunt-stack/)
• .claude/skills/*/SKILL.md
• .claude/skills/*/references/*.md

Changed mode: Analyze branch diff:

1. Get changed files: git diff --name-only origin/main...HEAD
2. Find directly modified Claude docs
3. Find docs that reference changed paths (smart detection)

Phase 2: Parallel Validation (Haiku Agents)

Spawn parallel Haiku agents organized by validation type:

Agent Group A: Path Validators (by directory)

One agent per major directory validates file/directory references:

• infrastructure/ paths
• kubernetes/ paths
• .taskfiles/ paths
• Root and other paths

Validation checks:

• File paths mentioned exist: [path/to/file](path/to/file)
• Directory paths exist: infrastructure/modules/
• Glob patterns return results: **/*.tf

Agent Group B: Code Reference Validators

Validate code-level claims:

• Function/class names exist where documented
• Line number references are approximately accurate (±10 lines)
• Variable names in examples exist in referenced files

Agent Group C: Command Validators

Validate command examples:

• task <name> commands exist in Taskfile
• CLI tools referenced are in Brewfile
• Command syntax is valid (dry-run where safe)

Agent Group D: Cross-Reference Validators

Validate documentation consistency:

• Skill references in CLAUDE.md match actual skills
• Table entries match actual directories
• Runbook references exist

Phase 3: Aggregation

Collect results from all agents into a unified report grouped by severity (CRITICAL / WARNING / INFO), showing doc path, line number, the broken reference, and a suggested fix. Count valid references. Present issues before proposing edits.

Phase 4: Opus Validation

Spawn Opus agent to:

1. Review aggregated findings
2. Prioritize by impact (breaking vs cosmetic)
3. Generate proposed edits for each issue
4. Present final change list for approval

Agent Responsibilities

Discovery Agent (Haiku): Use discover-claude-docs.sh (or discover-claude-docs.sh --changed) to enumerate files. For each file, run extract-references.sh to get path refs, dir refs, commands, skill refs, and CLI tools as structured JSON. Scripts are in .claude/skills/sync-claude/scripts/.

Path Validator Agents (Haiku, one per directory domain): Verify each path ref exists using Glob/Read. For misses, fuzzy-search for renames. Return valid/invalid lists with suggestions.

Command Validator Agent (Haiku): Verify task <name> commands exist in Taskfile.yaml or .taskfiles/*. Verify CLI tools appear in Brewfile. kubectl/git: syntax check only. Return valid/invalid with suggestions.

Change Impact Analyzer (Haiku, changed mode only): Parse git diff --name-only origin/main...HEAD. Find docs directly modified. Search all docs for references to changed paths/directories. Return impacted doc list.

Opus Validator Agent: Deduplicate all Haiku findings. Categorize severity (CRITICAL: broken refs users will hit; WARNING: outdated but functional; INFO: cosmetic). Generate proposed Edit operations (old_string/new_string). Present for user approval before applying.

Mode Selection Logic

IF user specifies mode:
  USE specified mode
ELSE IF on main branch:
  USE full mode
ELSE IF branch has commits ahead of origin/main:
  USE changed mode
ELSE:
  USE full mode (no changes to analyze)

Exclusions

Always exclude from scanning:

• .terragrunt-cache/
• .terragrunt-stack/
• node_modules/
• .git/
• *.rendered/

Error Handling

If an agent fails:

1. Log the failure with context
2. Continue with other agents
3. Mark affected validations as "INCOMPLETE"
4. Include in final report for manual review