research-general
Pass
Audited by Gen Agent Trust Hub on May 12, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No malicious intent or suspicious patterns detected. The skill follows best practices for information gathering.
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests untrusted data from the web using WebSearch and WebFetch. However, it includes defensive instructions for source evaluation and multi-source verification.
- Ingestion points: External data enters through WebSearch and WebFetch tools (referenced in SKILL.md).
- Boundary markers: The skill uses section headers but lacks explicit instruction/data delimiters for fetched content.
- Capability inventory: The skill uses Grep, Glob, Read for local file access and writes reports to the .workflow/research/ directory. It does not have access to sensitive credentials or shell command execution tools.
- Sanitization: The skill relies on a manual evaluation framework defined in references/source-evaluation.md and references/cross-validation.md rather than automated technical sanitization.
Audit Metadata