vibe-debug
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill architecture is designed to process external, potentially untrusted data to inform state-changing operations.
- Ingestion points: The agent is instructed to read user-supplied reports, repository files, and results from delegated investigators (sub-agents or proxies) to identify root causes and verify fixes (SKILL.md, Delegated Findings).
- Boundary markers: The instructions do not mandate specific technical delimiters (e.g., XML tags or unique markers) for external inputs, relying instead on the agent's ability to distinguish 'claims' from 'proof'.
- Capability inventory: The skill permits the agent to perform file edits and execute shell commands within a declared 'state-changing' boundary (SKILL.md, Effect And Write Boundaries).
- Sanitization: The skill lacks automated sanitization or escaping of external content, opting for manual 'verification with evidence' and 'self-review' gates to mitigate risks before closure.
Audit Metadata