Lark Toolkit

Lark Toolkit

Prerequisites & Security

This skill is a documentation-only reference guide. It contains no executable code that accesses credentials automatically.

Required credentials (user-provided, never bundled):

• Lark App ID (app_id) — from Lark Developer Console
• Lark App Secret (app_secret) — from the same console

How credentials are used:

• All API examples in SKILL.md use placeholders (<APP_ID>, <APP_SECRET>, CHAT_ID, etc.) — no real secrets
• The scripts/get_token.sh helper obtains a temporary tenant_access_token from Lark's auth API. It reads credentials from (in order):
  1. Command-line arguments
  2. LARK_APP_ID / LARK_APP_SECRET environment variables
  3. ~/.openclaw/openclaw.json (standard OpenClaw config, path channels.lark.accounts.default.appId/appSecret)
• The script prints the config source to stderr when falling back to the config file
• No credentials are hardcoded, cached to disk, logged, or transmitted beyond the single Lark auth API call
• The token is exported as LARK_TOKEN env var for subsequent commands in the same shell session

Three Access Paths

Rule: claw-lark first → MCP second → direct API as fallback.

Authentication (Direct API)

TOKEN=$(curl -s -X POST 'https://open.larksuite.com/open-apis/auth/v3/tenant_access_token/internal' \
  -H 'Content-Type: application/json' \
  -d '{"app_id":"<APP_ID>","app_secret":"<APP_SECRET>"}' \
  | python3 -c "import sys,json; print(json.load(sys.stdin)['tenant_access_token'])")

Or use the helper: bash scripts/get_token.sh

Token validity: ~2 hours. Cache it.

API Base URLs

⚠️ Lark ≠ Feishu. Always confirm which platform the tenant uses.

Common API Patterns

Send a Message

curl -X POST "https://open.larksuite.com/open-apis/im/v1/messages?receive_id_type=chat_id" \
  -H "Authorization: Bearer $TOKEN" -H "Content-Type: application/json" \
  -d '{"receive_id":"CHAT_ID","msg_type":"text","content":"{\"text\":\"hello\"}"}'

Reply in Thread

curl -X POST "https://open.larksuite.com/open-apis/im/v1/messages/MSG_ID/reply" \
  -H "Authorization: Bearer $TOKEN" -H "Content-Type: application/json" \
  -d '{"msg_type":"text","content":"{\"text\":\"reply\"}","reply_in_thread":true}'

Add Reaction

curl -X POST "https://open.larksuite.com/open-apis/im/v1/messages/MSG_ID/reactions" \
  -H "Authorization: Bearer $TOKEN" -H "Content-Type: application/json" \
  -d '{"reaction_type":{"emoji_type":"THUMBSUP"}}'

Emoji types: THUMBSUP HEART LAUGH OK COOL FINGERHEART SMILE JIAYOU

List Department Users (MCP gap — direct API only)

# List root departments
curl -s -H "Authorization: Bearer $TOKEN" \
  'https://open.larksuite.com/open-apis/contact/v3/departments?parent_department_id=0&page_size=50&fetch_child=true'

# List users in a department
curl -s -H "Authorization: Bearer $TOKEN" \
  'https://open.larksuite.com/open-apis/contact/v3/users?department_id=<DEPT_ID>&page_size=50'

Key fields: name, open_id, employee_type (1=regular, 2=intern), department_ids

Read Chat History

curl -s -H "Authorization: Bearer $TOKEN" \
  'https://open.larksuite.com/open-apis/im/v1/messages?container_id_type=chat&container_id=<CHAT_ID>&page_size=20&sort_type=ByCreateTimeDesc'

Add Bot to Group

curl -X POST "https://open.larksuite.com/open-apis/im/v1/chats/<CHAT_ID>/members?member_id_type=app_id" \
  -H "Authorization: Bearer $TOKEN" -H "Content-Type: application/json" \
  -d '{"id_list":["<BOT_APP_ID>"]}'

MCP Tools (38 available)

mcporter call lark-mcp.<tool_name> key=value

Full catalog with parameters: references/mcp-tools.md

MCP Coverage

MCP Gaps (use direct API)

• List users by department — GET /contact/v3/users?department_id=
• List departments — GET /contact/v3/departments
• Add/remove group members — POST /im/v1/chats/{chat_id}/members
• Send reactions — POST /im/v1/messages/{msg_id}/reactions
• Upload images/files — POST /im/v1/images / POST /im/v1/files

Pagination

Most list APIs use cursor-based pagination:

?page_size=50&page_token=<token_from_previous_response>

Check has_more in response.

Error Handling

Critical Pitfalls

1. Lark ≠ Feishu — International uses open.larksuite.com, China uses open.feishu.cn
2. open_id is per-app — Same user has different open_id across different Lark apps
3. Webhook 5s timeout — Return 200 immediately, process async
4. Event dedup — Use event_id (Lark retries up to 3x)
5. Bot-to-bot blind spot — Lark does NOT push Bot A's messages to Bot B's webhook
6. Publishing required — Permission/event changes only take effect after publishing a new app version
7. ngrok IPv6 trap — Use 127.0.0.1:PORT not localhost:PORT in ngrok config
8. ngrok free domain — Returns interstitial HTML that Lark rejects. Use paid domain.

Detailed References

• Bot setup playbook: references/bot-setup.md
• API reference: references/api-reference.md
• MCP tools catalog: references/mcp-tools.md
• Webhook & tunnel: references/webhook-setup.md
• Troubleshooting: references/troubleshooting.md
• Permissions: references/permissions.md