The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/sync_doc_to_bitable.py uses the Python subprocess module to call lark-cli for fetching documents and writing Bitable records. These calls use argument lists rather than shell strings, preventing command injection.
[DATA_EXFILTRATION]: The skill reads data from Feishu Docs and writes to Feishu Bitable. This data flow is contained within the official platform services and uses the user's local authentication. No unauthorized data transmission to external servers was detected.
[PROMPT_INJECTION]: The skill processes document content which serves as an untrusted input source for field extraction.
Ingestion points: Document text is fetched in scripts/sync_doc_to_bitable.py and parsed via parse_doc.
Boundary markers: None; the script relies on Markdown headers for extraction.
Capability inventory: The script can modify Bitable tables through the lark-cli api command.
Sanitization: Extracted fields are used as-is in API calls. This represents a surface for indirect injection where document content controls Bitable field values, which is the intended functionality of the skill.

ai-news-bitable-archive