feishu-agent

You can perform Feishu Bitable CRUD operations by issuing natural language commands like `/feishu list <table> where <condition>`. The agent parses the URL, validates the schema, and applies upsert-first logic to safely create, update, or delete records.

URL parsing for Feishu Bitable automation extracts app_token, table_id, and optional record_id directly from Wiki, Base, or Record URLs. Results are cached in EverMem with a 7-day TTL to accelerate subsequent automated operations.

Yes, natural language updates use upsert-first logic to query for duplicates before patching or creating records. Destructive actions like delete require explicit confirmation, preventing accidental data loss during automation.

No, the agent automatically maps human-readable permission descriptions to standard lark-cli authorization scopes. If permission is denied, it prompts you to re-authorize without requiring manual scope configuration.

Deleting Feishu Bitable records triggers a guarded write mechanism that requires explicit user confirmation before executing the destructive action. This safety boundary prevents accidental data deletion during automated workflows.

Repeating schema queries happens without caching. This agent introspects fields, types, and option enumerations per base, caching the schema in EverMem to eliminate redundant lookups and accelerate persistent data workflows.