diagnosing-bugs
Pass
Audited by Gen Agent Trust Hub on Sep 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructions specify processing external, potentially untrusted data sources to construct reproduction loops.
- Ingestion points: Phase 1 (Building a feedback loop) and the failure recovery section in SKILL.md mention reading 'captured artifacts', 'network requests/payloads', 'event logs', 'HAR files', and 'log dumps' from the environment.
- Boundary markers: The instructions do not specify the use of delimiters or 'ignore' warnings for the text contained within ingested artifacts.
- Capability inventory: The skill encourages the agent to create and execute shell commands, perform network requests via curl, and use browser automation tools like Playwright and Puppeteer.
- Sanitization: Instructions focus exclusively on redacting sensitive environment variables and credentials (using ) rather than sanitizing content for potential malicious instructions.
Audit Metadata