spec-executor
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to identify and execute instructions contained within SPEC READY blocks found in conversation history or referenced spec artifacts.
- Ingestion points: In SKILL.md, the instructions mandate finding the latest SPEC READY block in inherited conversation history and reading referenced spec artifacts to drive implementation.
- Boundary markers: The skill uses the SPEC READY string as an identifier but does not specify delimiters or security instructions to prevent the agent from obeying malicious commands embedded within those specs.
- Capability inventory: In SKILL.md, the skill is granted the capability to modify the repository, execute tests and repository-level checks, and potentially perform external actions like pushing code or triggering deployments if authorized by the spec content.
- Sanitization: The skill includes an explicit instruction to redact credentials, tokens, cookies, and personal data from the final execution receipt before it leaves the thread in SKILL.md.
Audit Metadata