diagnosing-bugs
Pass
Audited by Gen Agent Trust Hub on Aug 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill methodology requires the agent to ingest and analyze external artifacts, which could potentially contain malicious instructions.\n
- Ingestion points:
SKILL.mdspecifies reading external data such as log dumps, HAR files, core dumps, and event traces provided by the user.\n - Boundary markers: The instructions lack explicit requirements for the agent to use delimiters or boundary markers when processing these external inputs.\n
- Capability inventory: The skill involves powerful capabilities including shell execution (
bash), network requests (curl), and browser automation (Playwright/Puppeteer) as defined in the Phase 1 feedback loop options.\n - Sanitization: There are no instructions for sanitizing or filtering the content of ingested artifacts before processing them.\n- [DYNAMIC_EXECUTION]: The skill facilitates the creation and execution of custom scripts and harnesses to reproduce bugs.\n
- Evidence:
SKILL.mdencourages building throwaway harnesses, CLI invocations, and usingscripts/hitl-loop.template.shfor structured reproduction loops.\n - Context: The dynamic generation of scripts is directed at local debugging and reproduction of reported issues, which is standard for this agent's purpose.
Audit Metadata