blocking-io-guard
Pass
Audited by Gen Agent Trust Hub on Jun 30, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the user to execute local project commands such as
uv run --project backend python scripts/scan_changed_blocking_io.pyandmake detect-blocking-io. These commands are standard for the project's development workflow and are used to identify potential performance regressions in the event loop. - [DATA_EXFILTRATION]: No network operations or access to sensitive local data (e.g., credentials, SSH keys) were identified. The skill's operations are confined to the local repository and the processing of code scan results.
- [PROMPT_INJECTION]: The instructions are purely technical and focus on developer workflows. There are no patterns suggesting attempts to override system prompts or bypass AI safety guardrails.
- [EXTERNAL_DOWNLOADS]: The skill does not involve downloading external scripts or packages. All referenced files and tools are part of the local repository structure.
Audit Metadata