NYC

google-docs

Warn

Audited by Gen Agent Trust Hub on Feb 19, 2026

Risk Level: MEDIUMDATA_EXFILTRATIONCOMMAND_EXECUTIONNO_CODE
Full Analysis
  • [Data Exposure & Exfiltration] (MEDIUM): The documentation states that OAuth tokens are refreshed via a cloud function. Since the source code for auth.py is missing, the destination and ownership of this function cannot be verified, which poses a risk of credential interception.\n- [Indirect Prompt Injection] (LOW): The skill reads untrusted data from external documents which could contain malicious instructions directed at the agent.\n
  • Ingestion points: scripts/docs.py get-text (extracts text from user-controlled Google Docs).\n
  • Boundary markers: Absent; no documented delimiters or instructions to ignore embedded instructions.\n
  • Capability inventory: scripts/docs.py can create and modify documents (append, insert, replace text).\n
  • Sanitization: Absent; no evidence of input validation or content filtering.\n- [Unverifiable Implementation] (MEDIUM): The core implementation scripts (auth.py and docs.py) are referenced in the documentation but not provided in the package, preventing a full security audit of command execution or network calls.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 19, 2026, 04:40 AM