diagnose
Pass
Audited by Gen Agent Trust Hub on May 16, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill outlines a disciplined, industry-standard workflow for bug diagnosis (reproduce, minimise, hypothesise, instrument, fix). All suggested actions, such as running tests, using curl for HTTP requests, and executing git bisect, are appropriate for a developer-oriented tool.
- [COMMAND_EXECUTION]: The skill encourages building a feedback loop by running local tests, CLI commands, and headless browser scripts. This is the intended primary function of the skill and occurs within the developer's local environment.
- [PROMPT_INJECTION]: The skill contains a surface for indirect prompt injection as it processes external debugging data.
- Ingestion points: Error messages, logs, captured traces, and responses from the hitl-loop.template.sh script.
- Boundary markers: None explicitly defined in the skill instructions.
- Capability inventory: Execution of shell scripts, CLI tools, and network requests (curl) within SKILL.md and the template script.
- Sanitization: No specific sanitization or validation logic for the ingested data is provided in the instructions.
Audit Metadata