interview-me
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from existing files and external sources to inform the interview process.
- Ingestion points: Step 1 (Read context first) and Step 3 (Verify before resolving) where the agent skims project files and external facts.
- Boundary markers: Absent; there are no instructions to delimit or ignore potentially malicious instructions embedded in the project-based or external content.
- Capability inventory: File reading for context skimming and file writing to the
meta/interviews/directory. - Sanitization: Absent; no explicit escaping, filtering, or validation is applied to external or project content before it is processed or summarized.
Audit Metadata