omm

omm doctor

Diagnose the omm install itself and its Ollama links — which binary is actually running, whether the package and source agree, whether the registry matches what Ollama can see — without changing anything.

01 / 06

Overview

Reach for doctor when something about omm itself seems off — the wrong version running, a command not found after an update, a model omm thinks is linked that Ollama doesn't actually see — and you want a structured answer instead of guessing. Every check is read-only: nothing it finds gets fixed automatically, and running it twice in a row never changes the second result.

02 / 06

Options

Every flag this command accepts, and what it defaults to when you leave it out.

  • --json—Default: off

    Print the same checks as structured JSON instead of a table.

03 / 06

Examples

From a plain search to something you'd put in a script.

The full diagnostic table.

$ omm doctor

The same checks as JSON.

$ omm doctor --json

Same diagnosis, without the progress output.

$ omm doctor --quiet

04 / 06

A real run

Real omm doctor capture, 2026-08-25, this dev machine — personal file paths generalized, every check name, status, and non-path detail (versions, commit, tag names) real and unedited, including the real WARN this run actually found (a git-editable install whose package metadata trails its source by a few versions).

06 / 06

If something goes wrong

Every message below is one this command actually prints. Find yours, read why it happened, then do the last line.

  1. Overall: FAIL (N pass, N warn, 1 fail)
    why
    doctor's own docstring is explicit about this: a WARN never fails the command, but a genuine FAIL (like a completely broken Ollama link) does.
    what to do
    Read the FAIL row's detail column for the specific check that failed, and address that — the table names exactly what's wrong.
    source
    src/omm/cli.py:2513-2514

Still stuck? Open an issue with the exact message you saw.

All commands