The Agent Skills Directory

[SAFE]: The skill implements a 'doctor' script (scripts/asr_doctor.py) that checks for the presence of necessary environment variables (like STEP_API_KEY and VOLC_ASR_TOKEN) using boolean checks. It explicitly avoids printing or logging the actual secret values.
[SAFE]: Instructions in SKILL.md emphasize user safety and billing transparency, requiring manual confirmation for large tasks and directing users to official console URLs for financial management.
[SAFE]: Command execution is limited to standard Python scripting for routing logic and system readiness checks. The use of subprocess.run in tests/test_contract.py is a standard practice for unit testing the routing script's output.
[SAFE]: Data processing is focused on local audio and video files, with no evidence of unauthorized network exfiltration or remote code execution. All external references target official documentation or billing portals of well-known service providers.

yichen-asr