The Agent Skills Directory

[SAFE]: No malicious patterns or security vulnerabilities were identified. The skill demonstrates defensive coding practices and significant security awareness.
[COMMAND_EXECUTION]: The script scripts/compile_route_plan.py uses subprocess.run to invoke a validation script (validate_route_plan.py). This is a legitimate architectural choice for modular validation of generated plans and does not present a security risk.
[EXTERNAL_DOWNLOADS]: The script scripts/model_preflight.py performs network probes using urllib.request. These probes are strictly limited to loopback/localhost addresses by the loopback_probe_url function to prevent SSRF and data exfiltration.
[INDIRECT_PROMPT_INJECTION]: The skill processes task instructions for sub-agents, creating a surface for indirect prompt injection. However, it mitigates this risk by using structured Task Packets with explicit boundary markers, intent definitions, and mandatory validation scripts that enforce schema consistency.

codex-model-routing-team