The handful of things that actually go wrong on first run, and how to clear them.

Most problems fall into one of a few buckets. Start here before opening an issue.

Jarvis won’t boot after install

Run the two diagnostic commands first:

python -m jarvis --check      # is the hardware/capability picture sane?
python -m jarvis --plugins    # did the providers actually register?

If --plugins shows an empty or partial registry, the editable install didn’t take. Re-run:

pip install -e . --no-deps

“No provider configured” / no answer

You need at least one brain key. Re-run the wizard:

python -m jarvis --wizard

Keys live in your OS credential manager, not in jarvis.toml. If you rotated a key, update it through the wizard — editing the TOML won’t help, because secrets aren’t stored there.

Jarvis can’t hear me

Confirm the microphone is the OS default input device, or set it explicitly in jarvis.toml.
Check that no other app holds exclusive access to the mic.
Verify the wake word fired: run with python -m jarvis --debug and watch the console.

Jarvis is silent (no speech back)

Confirm an output device is selected and not muted.
Run --debug and look for text-to-speech errors in the log.

It runs but won’t act on my computer

Computer-use actions are gated by the risk-tier system. An action sitting at the ask tier is waiting for your approval; an action at block is refused by policy. Check which tier the action resolves to.

Headless server has no audio

That’s expected — a headless box has no microphone or speakers. Reach the full experience through the browser UI over WebSocket, or through a channel adapter (Telegram, Discord, SMS, webhook). See installation → headless.

Still stuck?

Read the source and open an issue on GitHub. Include the output of python -m jarvis --check.

Troubleshooting