internal-comms

Pass

Audited by Gen Agent Trust Hub on Mar 10, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it is instructed to read and summarize data from external, user-controlled sources such as Slack messages, Google Drive documents, and emails. An attacker could place malicious instructions within these sources (e.g., a Slack message or a shared document) that the agent might inadvertently follow when gathering context for reports or FAQs.
  • Ingestion points: The skill reads from Slack channels, Google Drive docs, Email threads, and Calendar events (specified in examples/3p-updates.md, examples/company-newsletter.md, and examples/faq-answers.md).
  • Boundary markers: The instructions do not include specific delimiters or 'ignore' instructions for the data being retrieved.
  • Capability inventory: The agent has the capability to read extensive organizational data and reformat it into summaries, newsletters, and FAQs.
  • Sanitization: There is no evidence of sanitization or validation of the content retrieved from these external platforms before it is processed by the LLM.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 10, 2026, 08:01 PM