autoresearch-skill
Pass
Audited by Gen Agent Trust Hub on Aug 4, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill operates entirely on local files within the repository or specified artifact directories. It uses standard Node.js/Bun file system modules to manage experiment logs and build a reporting dashboard. The implementation follows best practices for local tool development, including atomic file writes and data sanitization.
- [COMMAND_EXECUTION]: The skill utilizes shell search and a local script (
render-dashboard.mjs) to automate the optimization workflow. These operations are conducted within the local environment and are part of the skill's documented primary function to facilitate skill research and evaluation. - [DATA_EXFILTRATION]: No network exfiltration or non-whitelisted domain access was detected. The skill specifically instructs the agent to gate credentials and sensitive data during the optimization process and treats retrieved content as evidence rather than authoritative instructions.
- [PROMPT_INJECTION]: The skill includes comprehensive routing rules and trigger examples (including Korean language support) to ensure it only activates for intended optimization tasks. It defines clear boundaries between the researcher agent and the target skill being processed to mitigate steering risks.
Audit Metadata