Module intel 

Intelligence wire types — the provider-neutral representation the agentic loop reasons over.

The loop builds a Request and consumes a Response without knowing which provider answered. The intel/openai.rs and intel/anthropic.rs adapters translate to/from the on-the-wire JSON dialects; a model lacking native tool-calling falls back to the JSON-action shape parsed in agentloop/action.rs. Because the neutral model lives here rather than in one provider’s struct, supporting another provider costs one adapter and no change to the loop — that is what keeps the adapter count from growing into the rest of the runtime.

Structs

Request

A request to the intelligence endpoint.

Response

A response from the intelligence endpoint, normalized across providers.

ToolCall

A model-requested tool invocation. arguments is already-parsed JSON.

ToolDef

A tool advertised to the model in the request tools field. Sourced from the scoped MCP tools/list plus agentd’s own self-tools.

Usage

Token accounting from one model call. The supervisor sums these into the run’s budget, and a child’s usage also counts against its parent’s.

Enums

Message

One conversation message. Assistant may carry tool calls; ToolResult feeds a tool’s output back into the conversation as the next observation.

StopReason

Why the model stopped — drives the loop’s branch (tool-use vs final) and the exhausted_tokens terminal status.