chat-to-roadmap
Pass
Audited by Gen Agent Trust Hub on Jun 14, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted user-provided chat transcripts.
- Ingestion points: Phase 1 instructions specify reading pasted transcripts or summaries provided by the user.
- Boundary markers: Absent. The instructions do not define delimiters or specific warnings to ignore instructions embedded within the processed transcript.
- Capability inventory: The skill has the capability to write a file (
roadmap.html) to the local filesystem and invoke other skills (e.g.,grill-with-docs). - Sanitization: Absent. No sanitization or validation of the transcript content is performed before the agent processes and extracts information from it.
Audit Metadata