# `Sycophant.Pipeline`

Orchestrates the full LLM request lifecycle.

The pipeline executes these steps in order:

1. **Model Resolution** - Resolves the model spec to provider metadata via LLMDB
2. **Parameter Validation** - Validates LLM parameters through Zoi schemas
3. **Credential Resolution** - Finds credentials (per-request > app config > env vars)
4. **Wire Encoding** - Converts the request into provider-specific JSON
5. **HTTP Transport** - Sends the request via Tesla (sync or streaming)
6. **Wire Decoding** - Parses the provider response back into Sycophant structs
7. **Tool Execution** - Auto-executes tool calls if tools have functions set
8. **Cost Enrichment** - Calculates token costs from LLMDB pricing data
9. **Response Validation** - Validates structured output against schema if provided
10. **Context Attachment** - Stores conversation state for continuation

# `call`

```elixir
@spec call(
  [Sycophant.Message.t()],
  keyword()
) :: {:ok, Sycophant.Response.t()} | {:error, Splode.Error.t()}
```

Executes a full LLM request pipeline: resolves model, validates params, encodes, transports, and decodes.

---

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