sota-finder
Pass
Audited by Gen Agent Trust Hub on Nov 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted research data from external URLs and papers, which serves as an ingestion point for indirect prompt injection attacks.
- Ingestion points: Data is retrieved from arbitrary external domains via the
fetch_source.pyscript and theWebFetchtool. - Boundary markers: Absent. The instructions do not define explicit delimiters or instructions for the agent to ignore embedded commands within the research material.
- Capability inventory: The agent has access to
Bashfor script execution,Writefor artifact generation, andWebSearch/WebFetchfor network access. - Sanitization: Present. The
fetch_source.pyscript uses aReadableHTMLParserto filter active content such as<script>and<style>tags. Thebuild_outputs.pyscript useshtml.escapeto prevent XSS in the generated HTML reports. - [COMMAND_EXECUTION]: The skill executes internal utility scripts through the
Bashtool. - Evidence: The
SKILL.mdfile instructs the agent to runpython3 "<skill-root>/scripts/fetch_source.py"andpython3 "<skill-root>/scripts/build_outputs.py". - Risk: Potential for command injection exists if the agent interpolates user-controlled URLs or paths into the shell command without proper escaping, although the scripts themselves use
argparsefor argument handling.
Audit Metadata