langchainrust

A LangChain-inspired Rust framework for building LLM applications.

What it solves: Build Agents, RAG, BM25 keyword search, Hybrid retrieval, LangGraph workflows, MCP tools, Guardrails, multi-agent Handoffs - all in pure Rust.

Core Features

Full documentation: 中文文档 | English

Architecture

┌─────────────────────────────────────────────────────────────┐
│                      langchainrust                           │
├─────────────────────────────────────────────────────────────┤
│  LLM Layer                                                   │
│  ├── OpenAIChat / OllamaChat                                 │
│  ├── DeepSeek / Moonshot / Zhipu / Qwen (OpenAI compatible) │
│  ├── AnthropicChat (Claude API) / GeminiChat                 │
│  ├── Function Calling (bind_tools) / Streaming (stream_chat)│
│  └── 多模态 Vision (ImageContent + human_with_image)        │
├─────────────────────────────────────────────────────────────┤
│  Embeddings Layer                                            │
│  ├── OpenAIEmbeddings / DeepSeekEmbeddings                   │
│  └── QwenEmbeddings / MockEmbeddings                         │
├─────────────────────────────────────────────────────────────┤
│  Agent Layer                                                 │
│  ├── ReActAgent / FunctionCallingAgent                      │
│  ├── Plan-Execute Agent (规划-执行-重规划)                   │
│  ├── Handoffs (多 Agent 交接) / Streaming Function Calling  │
│  ├── GuardedAgent (Guardrails 安全护栏)                     │
│  ├── AgentExecutor                                          │
│  └── LangGraph (StateGraph, Subgraph, Parallel)             │
├─────────────────────────────────────────────────────────────┤
│  MCP Layer                                                   │
│  └── MCPClient (Stdio + SSE) -> MCPToolAdapter -> BaseTool   │
├─────────────────────────────────────────────────────────────┤
│  Retrieval Layer                                             │
│  ├── RAG (TextSplitter, VectorStore)                        │
│  ├── BM25 (Keyword Search, AutoMerging)                     │
│  ├── Hybrid (BM25 + Vector, RRF Fusion)                     │
│  ├── HyDE / MultiQuery / Reranking                          │
│  └── Loaders (Text/JSON/MD/PDF/CSV/HTML)                    │
├─────────────────────────────────────────────────────────────┤
│  Storage Layer                                               │
│  ├── Vector DB (InMemory, Qdrant, MongoDB, ChromaDB,        │
│  │              Redis, SQLite, PGVector, Pinecone)          │
│  └── Sessions (SessionManager + SessionStore)               │
├─────────────────────────────────────────────────────────────┤
│  Utility Layer                                               │
│  ├── Memory (Buffer, Window, Summary, SummaryBuffer)        │
│  ├── Chains (LLMChain, SequentialChain, RetrievalQA, ...)   │
│  ├── Prompts (PromptTemplate, ChatPromptTemplate, FewShot)  │
│  ├── Tools (Calculator, DateTime, URLFetch, HTTP/File/SQL)  │
│  ├── Output Parsers                                         │
│  ├── Token Counter (Tiktoken + Cost Tracking)               │
│  ├── LLM Cache                                              │
│  └── Callbacks (LangSmith, StdOut, FileHandler)             │
└─────────────────────────────────────────────────────────────┘

What's New in 0.3.0

• MCP 协议: 连接任意 MCP Server(stdio/SSE),工具自动适配为 BaseTool 供 Agent 调用
• 多模态 Vision: ImageContent + Message::human_with_image,OpenAI / Ollama 均支持
• Sessions 会话管理: SessionManager + 可插拔 SessionStore,多轮对话生命周期
• Token 计数器: TiktokenCounter + TokenTrackingLLM 用量统计 + ModelPricing 成本估算
• Guardrails 安全护栏: 输入/输出验证,SensitiveInfo / ForbiddenWords / MaxLength,GuardedAgent
• Plan-Execute Agent: 规划 → 执行 → 失败重规划(PlanExecuteAgent)
• Handoffs 多 Agent 交接: HandoffManager + HandoffTool,主 Agent 委托专业 Agent
• Streaming Tool Calls: StreamingFunctionCallingAgent 流式输出 + 工具调用事件
• 扩展工具: HTTPTool / FileTool(沙箱)/ SQLTool(只读)
• PGVector / Pinecone: 新增两个向量库后端
• HTML Loader: 去标签/脚本/样式,提取纯文本

详见 Usage Guide(中文)。

Installation

[dependencies]

langchainrust = "0.3.0"

tokio = { version = "1.0", features = ["full"] }



# Optional features

langchainrust = { version = "0.3.0", features = ["mongodb-persistence"] }  # MongoDB storage

langchainrust = { version = "0.3.0", features = ["qdrant-integration"] }    # Qdrant vector DB

langchainrust = { version = "0.3.0", features = ["redis-storage"] }         # Redis storage

langchainrust = { version = "0.3.0", features = ["sqlite-storage"] }        # SQLite storage (+ SQLTool)

langchainrust = { version = "0.3.0", features = ["pgvector-storage"] }      # PGVector (需自配 sqlx/pgvector 依赖)

# PineconeStore 无需 feature,默认可用(reqwest HTTP API)

Quick Start

use langchainrust::{OpenAIChat, OpenAIConfig, BaseChatModel};
use langchainrust::schema::Message;

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    let config = OpenAIConfig {
        api_key: std::env::var("OPENAI_API_KEY")?,
        base_url: "https://api.openai.com/v1".to_string(),
        model: "gpt-3.5-turbo".to_string(),
        ..Default::default()
    };
    
    let llm = OpenAIChat::new(config);
    
    let response = llm.chat(vec![
        Message::system("You are a helpful assistant."),
        Message::human("What is Rust?"),
    ], None).await?;
    
    println!("{}", response.content);
    Ok(())
}

Multi-Provider Support

use langchainrust::{
    DeepSeekChat, MoonshotChat, ZhipuChat, QwenChat,
    AnthropicChat, OllamaChat,
};

let deepseek = DeepSeekChat::from_env();
let moonshot = MoonshotChat::with_model("moonshot-v1-128k");
let claude = AnthropicChat::from_env();
let ollama = OllamaChat::new("llama3.2");

BM25 Keyword Search

use langchainrust::{BM25Retriever, Document};

let mut retriever = BM25Retriever::new();

retriever.add_documents_sync(vec![
    Document::new("Rust is a systems programming language"),
    Document::new("Python is a scripting language"),
]);

let results = retriever.search("systems programming", 3);

for result in results {
    println!("Document: {}", result.document.content);
    println!("Score: {}", result.score);
}

More examples in Usage Guide (中文).

Examples

examples/ 目录提供 12 个可运行示例,覆盖核心功能:

需要 API Key 的示例从环境变量读取:

export OPENAI_API_KEY="your-key"

cargo run --example basic_chat

无需 API Key 的示例(BM25 / LangGraph / Memory / Loader)可直接运行,适合快速体验。

Documentation

Testing

cargo test

Contributing

Contributions welcome! See CONTRIBUTING.md.

License

MIT or Apache-2.0, at your option.