think-research-framework
Pass
Audited by Gen Agent Trust Hub on Oct 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill researches external thinking frameworks by delegating to a subagent that performs web research, creating an entry point for untrusted content to influence the agent's output.
- Ingestion points: User-supplied framework names and external content retrieved from the web via the subagent (SKILL.md).
- Boundary markers: No delimiters or explicit instructions to ignore embedded commands in retrieved content are present in the skill definition.
- Capability inventory: The skill writes dossiers to the local filesystem in a designated staging directory (frameworks/_proposed/) and executes a local validation script (scripts/check-proposed-entry.mjs).
- Sanitization: No explicit sanitization or validation steps for web-retrieved content are mentioned before the data is processed or written to the filesystem.
Audit Metadata