inspector-review-update
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from OpenSpec files which could contain malicious instructions designed to manipulate the review or patching process.
- Ingestion points: The skill reads files like proposal.md, design.md, and tasks.md from the target change directory under openspec/changes/.
- Boundary markers: No explicit delimiters or boundary instructions are identified when file content is inlined into specialist subagent prompts.
- Capability inventory: The skill utilizes Read, Write, and Edit capabilities to modify OpenSpec files and dispatches specialized subagents to perform codebase exploration and analysis.
- Sanitization: No sanitization or content validation of the ingested files is performed before processing.
Audit Metadata