implement-spec
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill coordinates subagents to process specifications and tickets from external sources, which act as untrusted ingestion points. These inputs could contain malicious prompts designed to manipulate the implementer or exploration agents.
- Ingestion points: Technical specifications and task tickets read by the agent in Step 1 of
SKILL.md. - Boundary markers: The skill instructions do not specify the use of delimiters or warnings to ignore instructions embedded within the provided specs.
- Capability inventory: The agents can create worktrees, modify code, create pull requests, and write research notes to the file system.
- Sanitization: No evidence of input validation or content filtering for the ingested task data.
- [NO_CODE]: The skill consists entirely of markdown instructions and YAML configuration files, with no executable scripts or binary files included in the package.
Audit Metadata