kf-g-agent-research-report-only-unless-approved
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill provides strict behavioral guidelines that prevent the agent from making destructive changes to the workspace during research-oriented tasks.
- [INDIRECT_PROMPT_INJECTION]: The skill allows for the reading of external documentation and local files, which are potential vectors for indirect prompt injection. The skill mitigates this by requiring user approval before any sensitive capabilities are used.
- Ingestion points: Local files and web documentation.
- Boundary markers: Not specified.
- Capability inventory: File system access and git operations are identified as available but restricted.
- Sanitization: No sanitization of ingested content is mentioned.
Audit Metadata