post-incident-review
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external, untrusted data sources which could contain malicious directives.
- Ingestion points: The process involves reading external incident reports, communication logs, and forensic artifacts as described in Sections 2 and 3.
- Boundary markers: Section 8 ('Prompt Injection Safety Notice') provides explicit, strong instructions to the agent to treat all analyzed content as data, never execute code found within it, and ignore any embedded instructions.
- Capability inventory: The frontmatter restricts the agent to
Read,Grep, andGlobtools, ensuring that even if an injection were attempted, the agent lacks the capabilities to perform network exfiltration, file writes, or command execution. - Sanitization: The skill relies on descriptive role boundaries and natural language instructions to ensure the agent maintains a passive analysis posture.
Audit Metadata