Add Provider

/add-provider — add a new model provider

Providers in Forge follow a uniform shape (see src/models/ollama.ts, src/models/openai.ts, src/models/anthropic.ts). Copy the shape, don't invent a new one.

Provider name: $ARGUMENTS.

1. Provider class — src/models/$ARGUMENTS.ts

• Export a class that matches the shape of existing providers (availability probe, chat/complete, token counting, streaming).
• Availability probe must be ~1.5s max. Long timeouts block the REPL cold-start budget.
• Use undici for HTTP (already a dependency). Do not add axios, node-fetch, or similar.
• Redact credentials via src/security/redact.ts before any log call.
• Map transport errors to ForgeRuntimeError with the right retryable flag and a recoveryHint where possible.

2. Registry — src/models/registry.ts

• Add the provider id to providerEnum.
• Register the class in initProviders.
• If the provider is local (Ollama-style, runs on the user's machine), add it to isLocalProvider.

3. Catalog — src/models/local-catalog.ts

• Add model id → family classification there.
• Do not write regexes inside your provider to detect model capabilities. The catalog is the single source of truth.

4. Router — src/models/router.ts

• Verify the router can reach the new provider via the standard routing rules. If routing needs a change, update the router and add a test in test/unit/adapter.test.ts.

5. Unit test

• Mock the transport (see test/unit/adapter.test.ts for the pattern).
• Cover: availability probe success/failure, chat/complete happy path, 429 retry logic, auth failure surfaces a non-retryable structured error.

6. Docs

• Update the provider table in docs/ARCHITECTURE.md §6.
• Bump the provider count in README.md "At a glance" if applicable.
• Add a one-line setup note to docs/SETUP.md.

Run /verify to close the loop.