google-drive-automation
Pass
Audited by Gen Agent Trust Hub on Oct 24, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from Google Drive which can contain malicious instructions. 1. Ingestion points: File names, contents, and metadata are read via
scripts/drive.py searchandscripts/drive.py list(referenced in SKILL.md) into the agent's context. 2. Boundary markers: No delimiters or instructions are provided in SKILL.md to isolate Drive content or warn the agent about embedded commands. 3. Capability inventory: The skill enables downloading files to the local system, uploading local files, and managing Drive contents through thedownload,upload, andtrashcommands (referenced in SKILL.md). 4. Sanitization: No sanitization or validation of the retrieved content is mentioned. - [COMMAND_EXECUTION]: The skill's core functionality relies on instructing the agent to execute shell commands invoking local Python scripts.
- [EXTERNAL_DOWNLOADS]: The skill depends on external scripts (
scripts/auth.py,scripts/drive.py) that are not bundled in the package, creating a dependency on unverified code. - [DATA_EXFILTRATION]: The documentation mentions using an external 'cloud function' for OAuth token refresh, which introduces a potential risk of credential interception if the endpoint is not an official Google service.
Audit Metadata