The Agent Skills Directory

[DATA_EXFILTRATION]: The skill logic in scripts/cli.py and instructions in SKILL.md involve accessing a sensitive credential file (credentials.json) for authentication with Google APIs. While required for the skill's purpose, this involves the handling and exposure of private service account keys.
[PROMPT_INJECTION]: The skill reads data from external Google Spreadsheets, which can be an ingestion point for malicious instructions intended to influence the agent (Indirect Prompt Injection).
Ingestion points: Data retrieved via the read command in scripts/cli.py.
Boundary markers: No delimiters or instructional warnings are used to separate spreadsheet content from the agent's prompt context.
Capability inventory: The skill provides extensive write, append, and clear capabilities for Google Sheets through the gspread library.
Sanitization: No filtering or validation is performed on the data fetched from the API before it is provided to the agent.

google-sheets