What causes "Model "${options.model}" not found" in can1357/oh-my-pi?

Run the model list command (or inspect modelRegistry) to see valid model identifiers and correct the string

can1357/oh-my-pi · error · Error

Model "${options.model}" not found

Error message

Model "${options.model}" not found

What it means

createCleanseAgentRuntime() resolves the requested model via resolveCliModel against a refreshed ModelRegistry. If resolution fails or yields no model, it throws with the registry's error or a generic 'Model "X" not found'. The model string must match a model known to the registry after auth/refresh.

Source

Thrown at packages/coding-agent/src/cleanse/agent.ts:85

	): Promise<CleanseAgentOutcome>;
	/** Steer late diagnostics into a running worker's chat; false when undeliverable. */
	followUp(worker: number, diagnostics: readonly CleanseDiagnostic[]): Promise<boolean>;
	close(result?: CleanseLoopResult): Promise<void>;
}

/** Resolve the requested model and create a fresh persisted cleanse session. */
export async function createCleanseAgentRuntime(options: {
	cwd?: string;
	model: string;
	hooks?: CleanseAgentHooks;
}): Promise<CleanseAgentRuntime> {
	const cwd = options.cwd ?? getProjectDir();
	const [settings, authStorage] = await Promise.all([Settings.init({ cwd }), discoverAuthStorage()]);
	const modelRegistry = new ModelRegistry(authStorage);
	await modelRegistry.refresh();
	const resolved = resolveCliModel({ cliModel: options.model, modelRegistry, settings });
	if (resolved.error || !resolved.model) {
		throw new Error(resolved.error ?? `Model "${options.model}" not found`);
	}
	const modelSelector = resolved.selector ?? formatModelString(resolved.model);
	const modelDisplay = formatModelString(resolved.model);
	const sessionManager = SessionManager.create(cwd);
	await sessionManager.setSessionName("Cleanse", "auto");
	sessionManager.appendCustomEntry("cleanse", {
		status: "running",
		model: modelDisplay,
		selector: options.model,
	});
	await sessionManager.ensureOnDisk();
	const sessionFile = sessionManager.getSessionFile();
	if (!sessionFile) throw new Error("Cleanse session could not be persisted");
	const eventBus = new EventBus();
	const toolSession: ToolSession = {
		cwd,
		hasUI: false,
		suppressSpawnAdvisory: true,

View on GitHub (pinned to 9690622007)

Solutions

Run the model list command (or inspect modelRegistry) to see valid model identifiers and correct the string
Use the full provider-qualified form (e.g. 'anthropic/claude-...') instead of a bare name
Authenticate the provider so refresh() discovers its models (check auth storage / `omp` auth setup)
Omit options.model if allowed, so the default model from settings is used

Example fix

// before
createCleanseAgentRuntime({ model: "claude-sonnet" });
// after
createCleanseAgentRuntime({ model: "anthropic/claude-sonnet-4-5" });

Defensive patterns

Strategy: validation

Validate before calling

const registry = new ModelRegistry(await discoverAuthStorage());
await registry.refresh();
const available = registry.getModels().some(m => formatModelString(m) === options.model || m.id === options.model);
if (!available) throw new Error(`model "${options.model}" unavailable; pick from registry`);

Try / catch

try {
  runtime = await createCleanseAgentRuntime({ model: userSelectedModel });
} catch (err) {
  if (err instanceof Error && err.message.includes('not found') && err.message.includes('Model')) {
    runtime = await createCleanseAgentRuntime({}); // fall back to default model
  } else throw err;
}

Prevention

Always use provider-qualified model strings from the registry's own listing
Refresh the registry after auth changes before resolving models
Don't hardcode model names in scripts; read them from config validated against the registry
Re-check after provider deprecations/renames

When it happens

Trigger: Calling the cleanse runtime (ensureRuntime) with options.model set to a model id/alias the ModelRegistry cannot resolve — unknown id, provider not authenticated, model removed upstream, or wrong format (missing provider prefix).

Common situations: Typo in --model flag or settings; using a model name from another tool's naming scheme; auth for the provider not configured so the model never appears in the registry; upstream renamed/deprecated a model.

Related errors

AI-assisted analysis of can1357/oh-my-pi@9690622007 (2026-08-31). Data as JSON: /api/errors/123a7121f2ee2c72. Report an issue: GitHub.