workflow-router

Pass

Audited by Gen Agent Trust Hub on Aug 12, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [SAFE]: The skill defines a robust routing logic for software development workflows. It prioritizes safety by requiring explicit user approval for any configuration changes and ensuring that existing instructions are not overwritten.
  • [COMMAND_EXECUTION]: The skill package includes a Python script (eval/evaluate.py) and a shell wrapper (eval/run-eval.sh) used to validate the skill's logic against defined scenarios. These scripts operate on local files provided within the skill directory and do not perform network operations.
  • [INDIRECT_PROMPT_INJECTION]: The skill defines a setup process that ingests data from local project files.
  • Ingestion points: Files including AGENTS.md, CLAUDE.md, ADRs, and GitHub templates (SKILL.md).
  • Boundary markers: None explicitly defined for the data ingestion phase.
  • Capability inventory: The skill proposes file modifications and routes tasks to downstream implementation tools like tdd and subagent-pipeline.
  • Sanitization: The skill relies on mandatory human review and approval ('Approval required: yes') rather than automated sanitization, which effectively mitigates the risk of processing malicious instructions from local files.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 12, 2026, 12:13 PM
Security Audit — agent-trust-hub — workflow-router