google-photos

Pass

Audited by Gen Agent Trust Hub on Feb 23, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill manages dependencies via the uv package manager, specifying standard libraries including google-api-python-client, google-auth-oauthlib, and requests. It also references a workspace setup script located in a sibling directory (~/.agents/skills/gmail/scripts/setup_workspace.py) for configuration.
  • [DATA_EXFILTRATION]: The skill accesses the Google Photos Library API to retrieve and manage user media. It stores OAuth 2.0 tokens locally in ~/.photos_credentials/token.json following standard security practices for desktop applications.
  • [PROMPT_INJECTION]: The skill has a surface for indirect prompt injection as it retrieves untrusted metadata (filenames and descriptions) from Google Photos.
  • Ingestion points: Metadata is ingested in scripts/google_photos.py via the list_items, search_items, and get_item functions.
  • Boundary markers: None; metadata is returned as standard JSON fields.
  • Capability inventory: The skill can write to the local filesystem (download_item), read from the local filesystem (upload_item), and perform network requests to Google APIs.
  • Sanitization: No sanitization or filtering is performed on the retrieved metadata before it is returned to the agent context.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 23, 2026, 09:36 AM