The Agent Skills Directory

[SAFE]: No malicious patterns such as credential theft, persistence, or obfuscation were detected. The skill uses a local binary and environment variables for legitimate spreadsheet management.
[COMMAND_EXECUTION]: Executes the local lark CLI and jq to process data from spreadsheets.
[PROMPT_INJECTION]: The skill creates an indirect prompt injection surface because it processes untrusted data from external spreadsheet sources.
Ingestion points: Data retrieved via lark sheet read.
Boundary markers: Not present in instructions.
Capability inventory: Command execution via lark and file writing via sheet download.
Sanitization: Data is handled as structured JSON without specific content filtering.
[DATA_EXFILTRATION]: Includes a feature to download spreadsheet attachments to the local filesystem, which involves interaction with the external Lark API.

sheets