1k-sentry
Pass
Audited by Gen Agent Trust Hub on Feb 18, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- PROMPT_INJECTION (LOW): Indirect Prompt Injection surface detected. The skill guides the agent to analyze external crash reports and provide code fixes.
- Ingestion points: Sentry dashboard and crash reports (untrusted external data entered at runtime).
- Boundary markers: No delimiters or 'ignore' instructions are specified to help the agent distinguish between data and instructions.
- Capability inventory: The skill implies the agent can modify the codebase to fix identified errors, providing a file-write vector.
- Sanitization: No sanitization or validation of the crash data is mentioned. Maliciously crafted error messages could potentially influence the agent's behavior during the fix phase.
Audit Metadata