internal-comms
Pass
Audited by Gen Agent Trust Hub on Aug 26, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) because it is designed to ingest and summarize large volumes of untrusted data from Slack, Email, and Google Drive to generate company-wide updates. An attacker could embed malicious instructions in a Slack message or a document which the agent might then follow when generating reports or newsletters. The skill lacks explicit boundary markers or sanitization instructions to mitigate these risks.
- Ingestion points: Processes messages from Slack channels, contents of emails, and documents from Google Drive (examples/3p-updates.md, examples/company-newsletter.md).
- Boundary markers: None found. No instructions tell the agent to ignore or delimit instructions found within the source material.
- Capability inventory: Accesses Slack, Email, Google Drive, and Calendar tools.
- Sanitization: None found. There are no instructions to validate, escape, or filter external content before interpolation.
Audit Metadata