diagnose

Pass

Audited by Gen Agent Trust Hub on Sep 16, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest untrustworthy data such as bug reports, logs, and network traces to construct and execute feedback loops. 1. Ingestion points: Input from users describing bugs or performance regressions in SKILL.md. 2. Boundary markers: Absent; there are no instructions to the agent to treat input as data rather than instructions. 3. Capability inventory: The skill utilizes subprocess execution for CLI tools, browser automation (Playwright/Puppeteer), and shell script execution from scripts/hitl-loop.template.sh. 4. Sanitization: Absent; the skill does not suggest any validation or sanitization of input before it is used in command-line arguments or script generation.
  • [DYNAMIC_EXECUTION]: The skill involves creating and running 'throwaway harnesses' and 'CLI invocations' based on templates or generated logic to reproduce bugs. This is a low-severity finding as it is part of the intended debugging workflow and uses local script generation.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 16, 2026, 12:13 PM
Security Audit — agent-trust-hub — diagnose