research-idea-validator
Pass
Audited by Gen Agent Trust Hub on May 5, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill implements a structured academic framework (FIVE+C) for evaluating research ideas without any signs of malicious intent or code obfuscation. The tools requested are consistent with the skill's stated purpose of research management and literature review.
- [COMMAND_EXECUTION]: The skill requests access to the Bash tool, which is used legitimately to manage project files, update research logs, and navigate the local file system (e.g., maintaining files in the memory/ and docs/ directories).
- [EXTERNAL_DOWNLOADS]: The skill utilizes WebSearch and WebFetch to perform legitimate research tasks such as identifying related work, checking for concurrent research, and validating novelty claims.
- [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection as it processes untrusted data from web sources and external documents during literature review steps.
- Ingestion points: Content retrieved via WebSearch and WebFetch tools, user-provided research notes, and external paper files used for comparison.
- Boundary markers: No explicit delimiters are used in the prompt instructions to isolate untrusted external content from the evaluation logic.
- Capability inventory: The skill possesses filesystem access via Bash/Write and network capabilities via WebFetch.
- Sanitization: External content is used directly for qualitative assessment without prior filtering or sanitization steps.
Audit Metadata