Anthropic API provider adapter.

Implements IAIProvider against the Anthropic Messages API.
Requires an API key in the configuration.

API docs: https://docs.anthropic.com/en/api/messages

No-op AI provider for tests and environments without AI.

Returns deterministic canned responses — no network calls,
no API keys needed. Used in test configs.

Ollama AI provider adapter.

Connects to a local Ollama instance (default: http://localhost:11434).
No API key required — offline-first by design.

Implements IAIProvider via the Ollama REST API:
POST /api/chat   — used for all completions
POST /api/generate — alternative for simple text generation

OpenAI API provider adapter.

Implements IAIProvider against the OpenAI Chat Completions API.
Requires an API key in the configuration.

Replicate provider adapter.

Implements IAIProvider against Replicate's predictions API, which hosts many
models — including Anthropic and Google ones — behind a single token.

Replicate has no OpenAI-compatible endpoint (`/v1/chat/completions` is a
404), so it cannot be driven by the openai adapter. Two further differences
shape this namespace:

- The documented flow is asynchronous: create a prediction, poll until it
  succeeds. The `Prefer: wait` header makes it synchronous, which is all
  Wagoe needs and avoids a polling loop entirely.
- `output` comes back as a list of string chunks rather than a string.

Input constraints are per-model rather than per-API. `anthropic/claude-4.5-haiku`
rejects `max_tokens` below 1024 with a 422 naming the field, so the failure
is reported with that detail instead of a bare status.