rpi-agent 0.1.1

Agent runtime + loop, Rust port of pi-agent-core core layer
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42

//! Mirrors `packages/agent/src` (core layer: agent + agent-loop + types + stream-fn).
//!
//! Provider-agnostic agent runtime. The only LLM boundary is [`StreamFn`]
//! (sync return → [`AssistantMessageEventStream`]); everything else talks in
//! [`AgentMessage`] and never touches a provider wire format.
//!
//! M2 scope per the plan: `AgentTool` trait, `Agent` + `AgentBuilder`,
//! `run_agent_loop` / `run_agent_loop_continue` free functions, events, hooks,
//! queues, abort. The critical invariants — tool-execution ordering
//! (end in completion order, tool-result `MessageEnd` in source/ordinal order),
//! truncate-fail, and late-update suppression — live in [`agent_loop`].

pub mod abort;
pub mod agent;
pub mod agent_loop;
pub mod agent_tool;
pub mod error;
pub mod events;
pub mod hooks;
pub mod message;
pub mod queue;
pub mod stream_fn;
pub mod types;

pub use abort::AbortHandle;
pub use agent::{Agent, AgentBuilder, AgentOptions};
pub use agent_loop::{run_agent_loop, run_agent_loop_continue, LoopOutcome, NewMessages};
pub use agent_tool::AgentTool;
pub use error::AgentError;
pub use events::{AgentEmitter, AgentEvent, CollectorEmitter};
pub use hooks::{
    AgentLoopConfig, AfterToolCall, BeforeToolCall, ConvertToLlm, GetApiKey, GetFollowUpMessages,
    GetSteeringMessages, PrepareNextTurn, ShouldStopAfterTurn, TransformContext,
};
pub use message::{AgentMessage, AgentMessageRole, CustomMessage};
pub use queue::PendingMessageQueue;
pub use stream_fn::{StreamFn, stream_fn};
pub use types::{
    AfterToolCallContext, AfterToolCallResult, AgentContext, AgentLoopTurnUpdate, AgentState,
    AgentToolResult, BeforeToolCallContext, BeforeToolCallResult, QueueMode,
    ShouldStopAfterTurnContext, TextContentOrImage, ToolExecutionMode, ToolResultPartial,
};