runner

Loop runner — composition root entry points for the conversation loop. Moved from core/loop/runner.gleam (Stage 3). Wires concrete service implementations into service bundles for the orchestrator.

Values

run

</>

pub fn run(
  client: client.Client,
  persona: String,
  history: List(message.Message),
  user_message: String,
  tool_defs: List(tool_def.ToolDefinition),
  max_rounds: Int,
  conn: sqlight.Connection,
  user_key: String,
  session_id: String,
) -> orchestrator.LoopResult

Run the conversation loop with the default tool registry (auto-deny approvals).

run_with

</>

pub fn run_with(
  client: client.Client,
  persona: String,
  history: List(message.Message),
  user_message: String,
  tool_defs: List(tool_def.ToolDefinition),
  max_rounds: Int,
  all_tools: List(tool.Tool),
  conn: sqlight.Connection,
  user_key: String,
  session_id: String,
  approval_handler: fn(tool.ApprovalRequest) -> tool.ApprovalResponse,
) -> orchestrator.LoopResult

Run the conversation loop with a custom tool list (e.g. approval-enabled bash).

run_with_chat_fn

</>

pub fn run_with_chat_fn(
  chat_fn: fn(
    client.Client,
    List(message.Message),
    List(tool_def.ToolDefinition),
  ) -> Result(response.ChatResponse, client.ClientError),
  client: client.Client,
  persona: String,
  history: List(message.Message),
  user_message: String,
  tool_defs: List(tool_def.ToolDefinition),
  max_rounds: Int,
  all_tools: List(tool.Tool),
  conn: sqlight.Connection,
  user_key: String,
  session_id: String,
) -> orchestrator.LoopResult

Run the conversation loop with a custom chat function (for testing, auto-deny approvals).