CLI Companion Reason

`ed companion reason`

Shows or changes how the companion reasons. Settings persist on the backend and hot-swap into the running service, so no restart or .env edit is needed; the environment remains the fallback for anything never set here.

Usage:

ed companion reason [show] [--json] [--endpoint <url>]
ed companion reason set [--provider <p>] [--model <m>] [--url <u>] [--api-key <k>]
                        [--json] [--endpoint <url>]
ed companion reason test [--json] [--endpoint <url>]

Options for ed companion reason set:

Name	Type / values	Default	What it does
`--provider`	supported provider id	unchanged	Which API shape to speak; `openai` covers any OpenAI-compatible server such as Ollama.
`--model`	model name	unchanged	Model to request; empty resets to the provider default.
`--url`	URL	unchanged	Base URL for the OpenAI-compatible provider.
`--api-key`	secret	unchanged	Stored in the backend's settings table, not the CLI config; empty clears it.

At least one setting option is required. Empty values remove that saved override and expose the provider's environment or built-in default again. --json returns the same masked settings object as show; it never echoes the key supplied on the command line.

ed companion reason show (also the bare default) prints the active provider, model, URL, whether a key is set with its last-four hint, and whether the reasoner is configured at all. --json shape: {provider, url, model, hasApiKey, apiKeyHint, configured, description}. The key itself is never returned.

ed companion reason test sends one tiny completion through the active provider and reports the round-trip: {ok, model, latencyMs} under --json, exit 1 with the provider's error when the backend reports failure. Failure to reach the companion API itself exits 4.