Skills
skills/koji98/agentflow/agentflow-run-debugging/Gen Agent Trust Hub

agentflow-run-debugging

Pass

Audited by Gen Agent Trust Hub on Apr 8, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill instructs the agent to analyze diagnostic log files and execution artifacts (such as stdout.log, stderr.log, and events.jsonl) which could potentially contain untrusted data that might influence agent behavior. This is inherent to the skill's primary function as a debugging tool.
  • Ingestion points: Diagnostic artifacts and execution log files referenced in SKILL.md and the artifact references.
  • Boundary markers: No specific boundary markers are defined for the inspection of log file content.
  • Capability inventory: The skill directs the use of the agentflow CLI for tasks like resume, validate, and compile based on the results of artifact inspection.
  • Sanitization: The instructions do not specify a requirement for sanitizing the content of log files before processing.
  • [COMMAND_EXECUTION]: The workflow involves the execution of local operational commands using the agentflow CLI toolset to manage and inspect graph contracts and run states.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 8, 2026, 04:24 PM