google-workspace
Pass
Audited by Gen Agent Trust Hub on Apr 20, 2026
Risk Level: SAFECREDENTIALS_UNSAFEDATA_EXFILTRATIONPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [CREDENTIALS_UNSAFE]: The skill's authentication scripts (scripts/auth.py and scripts/gmail/auth.py) reference and manage sensitive credential files including credentials.json, token.json, and service-account.json. While these are required for Google API access, they represent a data exposure risk if not handled correctly by the user or agent.
- [DATA_EXFILTRATION]: The skill provides powerful capabilities for data movement, including sending emails with attachments (scripts/gmail/send_email.py) and uploading/sharing files in Google Drive (scripts/drive/drive_files.py). These tools could be abused to exfiltrate sensitive data from the user's environment to external recipients.
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it retrieves and parses untrusted data from external sources (Gmail messages and Drive files).
- Ingestion points: scripts/gmail/parse_message.py and scripts/gmail/list_messages.py retrieve external message content; scripts/drive/drive_files.py retrieves file metadata and content.
- Boundary markers: Absent. The scripts do not implement delimiters or warnings to ignore instructions embedded in the processed data.
- Capability inventory: The skill includes broad capabilities for network operations (Google API calls), file modification, and message deletion across Gmail, Sheets, Docs, and Drive.
- Sanitization: Minimal. scripts/gmail/parse_message.py contains a basic regex for stripping HTML tags from email bodies, but no robust validation or instruction filtering is implemented.
- [EXTERNAL_DOWNLOADS]: The skill instructs the installation of official Google API client libraries (google-api-python-client, google-auth-httplib2, google-auth-oauthlib) from standard package registries.
Audit Metadata