insane-research-query
Pass
Audited by Gen Agent Trust Hub on Aug 1, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill acts as a structured interface for research planning. It uses local references for schema validation and example data, and its file operations are limited to saving user-confirmed research queries in a specific subdirectory.
- [INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection because it ingests untrusted user input (research topics) and incorporates it into a markdown brief and a JSON file. However, its capabilities are low-risk (local file writing to a specific path) and it serves a legitimate utility.
- Ingestion points: User input in Phase 1 (Discovery) and Phase 2 (Scoping) in
SKILL.md. - Boundary markers: None explicitly present in the instructional prompts.
- Capability inventory: File-writing to
RESEARCH/queries/documented in Phase 4 ofSKILL.md. - Sanitization: No explicit filtering or sanitization of user-provided research strings.
Audit Metadata