diagnose

Pass

Audited by Gen Agent Trust Hub on Jul 27, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill follows a standard software engineering diagnosis loop. The instructions and accompanying template script are legitimate and serve the intended purpose of debugging codebases.
  • [INDIRECT_PROMPT_INJECTION]: The skill involves processing potentially untrusted data from external sources (e.g., logs, HAR files, core dumps) as described in Phase 1 and 2 of SKILL.md. This represents a known surface for indirect prompt injection, but it is inherent to the diagnostic function. The evidence chain includes:
  • Ingestion points: External artifacts like HAR files, log dumps, and core dumps are loaded into the agent's context during Phase 1.
  • Boundary markers: The skill does not explicitly define delimiters for this external content, though the workflow is highly structured.
  • Capability inventory: The skill uses capabilities such as reading files, executing shell commands (e.g., git bisect, curl), and running test scripts.
  • Sanitization: There are no explicit instructions to sanitize or validate the content of external artifacts before processing.
  • [COMMAND_EXECUTION]: The skill guides the agent to use CLI tools (git, curl, wget) and run locally generated test scripts or the provided HITL bash template. These actions are standard for a debugging agent and are performed within the context of reproducing issues.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 27, 2026, 12:23 AM