L1: Public Document Sync

L1: Public Document Sync

Extraction Commands

Run these to collect current facts from code:

# Hook counts (per adapter — count event registrations, not top-level keys)
node -e "const h=JSON.parse(require('fs').readFileSync('platform/adapters/claude-code/hooks/hooks.json','utf8')).hooks; let c=0; for(const k in h) c+=h[k].length; console.log('claude-code:',c)"
node -e "const h=JSON.parse(require('fs').readFileSync('platform/adapters/gemini/hooks/hooks.json','utf8')).hooks; let c=0; for(const k in h) c+=h[k].length; console.log('gemini:',c)"
node -e "const h=JSON.parse(require('fs').readFileSync('platform/adapters/codex/hooks/hooks.json','utf8')).hooks; let c=0; for(const k in h) c+=h[k].length; console.log('codex:',c)"

# Shared module count
ls platform/adapters/shared/*.mjs | wc -l

# MCP tool count (name: entries in TOOLS array)
grep -c '"name":' platform/core/tools/mcp-server.mjs

# Test count
npm test 2>&1 | grep -oP 'tests \K\d+'

# Agent count (per adapter)
ls platform/adapters/claude-code/agents/*.md 2>/dev/null | wc -l
ls platform/adapters/gemini/agents/*.md 2>/dev/null | wc -l
ls platform/adapters/codex/agents/*.md 2>/dev/null | wc -l

# Skill count (per adapter + shared)
ls platform/skills/*/SKILL.md | wc -l
ls platform/adapters/claude-code/skills/*/SKILL.md | wc -l
ls platform/adapters/gemini/skills/*/SKILL.md | wc -l
ls platform/adapters/codex/skills/*/SKILL.md | wc -l

# Command count (per adapter)
ls platform/adapters/claude-code/commands/*.md 2>/dev/null | wc -l
ls platform/adapters/gemini/commands/*.toml 2>/dev/null | wc -l

# Version
node -e "console.log(JSON.parse(require('fs').readFileSync('package.json','utf8')).version)"

# Trigger factor count
grep -cP '^\s*//\s*\d+\.' platform/providers/trigger.ts

# Stagnation pattern count
grep -c 'pattern:' platform/bus/stagnation.ts 2>/dev/null

# Event type count
grep -cP "'" platform/bus/events.ts

# Language count + fragments
ls -d platform/core/languages/*/spec.mjs | wc -l
for lang in platform/core/languages/*/; do echo "$(basename $lang): $(ls $lang/spec.*.mjs 2>/dev/null | wc -l) fragments"; done

Target Files

File	Facts to Check
README.md	Version, hook counts, tool count, test count, module count, adapter table
README.ko.md	Same facts, Korean version — numbers must match EN
docs/README.md	Module map, hook counts, tool count, adapter table
docs/ko-KR/README.md	Same, Korean
docs/AGENTS.md	Trigger factors, tool table, stagnation patterns, language count
docs/ko-KR/AGENTS.md	Same, Korean
docs/TOOLS.md	One ## section per MCP tool
docs/ko-KR/TOOLS.md	Same sections, Korean

Matching Rules

Numeric Mismatch Detection

Search documents for these patterns and compare against extracted values:

(\d+)\s*(hooks?|훅)              → per-adapter or total hook count
(\d+)\s*(tools?|도구)            → MCP tool count
(\d+)\s*(tests?|테스트)          → test count
(\d+)\s*(modules?|모듈)          → shared module count
(\d+)\s*(agents?|에이전트)       → agent count
(\d+)\s*(skills?|스킬)           → skill count
(\d+)\s*(commands?|커맨드|명령)   → command count
(\d+)\s*(events?|이벤트)         → event type count
(\d+).*(factor|팩터)             → trigger factor count
(\d+).*(pattern|패턴)            → stagnation pattern count (within stagnation section)
(\d+).*(languages?|언어)         → language count
(\d+).*(fragments?|프래그먼트)    → fragment count

Context matters: verify the number refers to the correct adapter/module before changing. "22 hooks" in a Claude Code section means Claude Code hooks, not total.

Section Parity (EN/KO)

Extract ## headers from each EN/KO document pair:

• EN has section, KO doesn't → add (match existing style)
• KO has section, EN doesn't → report only (may be intentional)

Adapter Table

All adapters with platform/adapters/<adapter>/hooks/hooks.json must have a row in the provider/adapter table:

• Claude Code
• Gemini
• Codex

Tool Sections (TOOLS.md)

Every tool name in platform/core/tools/mcp-server.mjs TOOLS array must have a ## tool_name section in TOOLS.md (both EN and KO).