reconcile-roadmap
Pass
Audited by Gen Agent Trust Hub on Jul 16, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its reliance on processing external documentation files.
- Ingestion points: The agent reads data from decisions.md, seed.md, proposal.md, and roadmap.md within the thread root directory (docs/threads/).
- Boundary markers: There are no defined boundary markers to delimit external content, nor are there instructions to the model to ignore instructions embedded within the ingested data.
- Capability inventory: The skill is authorized to read multiple files and perform file-write operations specifically on roadmap.md.
- Sanitization: The instructions do not specify any sanitization, escaping, or validation of the content extracted from the authority files before it is interpolated into the target roadmap artifact.
Audit Metadata