# `LlmCall.Capabilities`
[🔗](https://github.com/gilbertwong96/llm_call/blob/main/lib/llm_call/capabilities.ex#L1)

What a completion protocol supports.

Capabilities are explicit so unsupported canonical fields can be rejected
before a transport sends anything, instead of being dropped silently.

`builtin_tools` declares provider-hosted tools (web search, grounding, server
tools): `false` rejects them, `:passthrough` forwards any name the caller
supplies, and a list forwards only those names. The library never interprets a
hosted tool's name; it is the provider's own identifier.

They are **protocol-level**: `capabilities/2` takes the request for signature
symmetry but does not inspect the model. A model-specific limit surfaces as a
provider error, not a fabricated local guess.

# `t`

```elixir
@type t() :: %LlmCall.Capabilities{
  builtin_tools: false | :passthrough | [String.t()],
  file_input: boolean(),
  image_input: boolean(),
  modalities: [LlmCall.modality()],
  parallel_tool_calls: boolean(),
  prompt_cache: boolean(),
  reasoning: LlmCall.Reasoning.Capabilities.t() | nil,
  stateful: boolean(),
  streaming: boolean(),
  streaming_tool_calls: boolean(),
  structured_output: :none | :json_object | :json_schema,
  tool_choice: boolean(),
  tools: boolean()
}
```

---

*Consult [api-reference.md](api-reference.md) for complete listing*
