gsd-project-researcher
Pass
Audited by Gen Agent Trust Hub on Mar 15, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill's operations and tool usage (Read, Write, Bash, WebFetch) are entirely consistent with its documented purpose of technology research and documentation. No suspicious or malicious behaviors were detected during analysis.
- [PROMPT_INJECTION]: The skill processes untrusted data from the web (via WebFetch) and external project files (PROJECT.md), which is an inherent surface for indirect prompt injection.
- Ingestion points: Data is retrieved from the internet via WebFetch and read from local project context files.
- Boundary markers: The skill utilizes structured XML-like delimiters (e.g.,
<research_type>,<question>) to isolate inputs, providing a basic mitigation against instruction confusion. - Capability inventory: The skill utilizes Bash, WebFetch, Read, and Write tools to gather data and generate research reports in Markdown format.
- Sanitization: No specific sanitization or filtering logic is defined for content retrieved from external sources, relying on the agent's synthesis logic to format the output.
Audit Metadata