debugging-patterns

Pass

Audited by Gen Agent Trust Hub on Mar 14, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill provides numerous examples of standard development commands, such as npm run build, npx tsc, and git bisect, to be executed via the bash tool as part of the debugging process.
  • [EXTERNAL_DOWNLOADS]: The documentation suggests the use of standard package managers like npm for installing dependencies when modules are missing.
  • [PROMPT_INJECTION]: The skill instructs the agent to analyze error messages and logs from external sources, which creates an attack surface for indirect prompt injection. Ingestion points: External error messages and stack traces. Boundary markers: None explicitly defined for untrusted log data. Capability inventory: Bash, Read, Grep, Glob, and LSP tools. Sanitization: No specific sanitization or filtering of external content is mentioned.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 14, 2026, 02:17 PM