Skip to content

Latest commit

 

History

History
31 lines (21 loc) · 1.22 KB

File metadata and controls

31 lines (21 loc) · 1.22 KB

Lessons

Known Setup

  • Bolo runs from ~/bolo and starts through start-bolo.command.
  • The current launcher builds target/release/bolo when needed and supervises the Rust runtime.
  • Logs go to /tmp/bolo.log.

Gotchas

  • Do not assume old Python entrypoints are authoritative. The launcher stops older bolo.py, hotkey.py, and overlay.py processes before starting the Rust binary.
  • macOS accessibility, microphone, and hotkey behavior should be verified after runtime changes.
  • Keep STT model names aligned with the public provider API docs. The default primary is Telnyx deepgram/nova-3, and the default rate-limit fallback is Telnyx openai/whisper-large-v3-turbo.
  • Do not hardcode experimental ASR models before they are available through the API Bolo actually calls. Expose model selection through BOLO_STT_MODEL and provider fallback through BOLO_STT_FALLBACKS so others can test new models without code changes.

Verification Commands

./start-bolo.command
cargo build --release
tail -n 100 /tmp/bolo.log

Do Not Break

  • Keep the login-item launcher path stable unless the Login Item is updated too.
  • Keep the lock files under /tmp coordinated with the launcher.

Last Updated

2026-05-18