weft-ai-language

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Weft AI Language

Weft AI 语言

Skill by ara.so — Daily 2026 Skills collection.

Weft is a programming language (implemented in Rust) for AI systems where LLMs, humans, APIs, databases, and agents are base language primitives. You wire nodes together, the compiler type-checks every connection, and the program runs with durable execution backed by Restate (survives crashes, supports multi-day human-in-the-loop pauses). A visual graph view is generated automatically from code.

由ara.so提供的技能 — 2026每日技能合集。

Weft是一款基于Rust实现的AI系统编程语言，其中LLM、人类、API、数据库和Agent都是语言的基础原语。你可以将节点连接在一起，编译器会检查每一个连接的类型，程序依托Restate实现持久化执行（可在崩溃后恢复，支持多日的人机交互暂停）。代码会自动生成可视化图形视图。

Installation & Setup

安装与配置

Prerequisites

前置要求

Docker (for PostgreSQL)
Node.js
macOS:
```
brew install bash
```
(Bash 4+ required)
Rust, Restate, and pnpm are auto-installed by
```
dev.sh
```

Docker（用于PostgreSQL）
Node.js
macOS：
```
brew install bash
```
（需Bash 4及以上版本）
Rust、Restate和pnpm会由
```
dev.sh
```
自动安装

Clone and Configure

克隆与配置

bash

git clone https://github.com/WeaveMindAI/weft.git
cd weft
cp .env.example .env

bash

git clone https://github.com/WeaveMindAI/weft.git
cd weft
cp .env.example .env

Edit .env — add your API keys

编辑.env — 添加你的API密钥

undefined

undefined

Environment Variables (

.env

)

环境变量（

.env

）

bash

OPENROUTER_API_KEY=     # Required for LLM nodes
TAVILY_API_KEY=         # Required for Web Search nodes
ELEVENLABS_API_KEY=     # Required for Speech-to-Text nodes
APOLLO_API_KEY=         # Required for Apollo enrichment nodes
DISCORD_BOT_TOKEN=      # Required for Discord nodes

All keys are optional at startup — missing keys surface as runtime errors only when the relevant node executes.

bash

OPENROUTER_API_KEY=     # LLM节点必填
TAVILY_API_KEY=         # 网页搜索节点必填
ELEVENLABS_API_KEY=     # 语音转文本节点必填
APOLLO_API_KEY=         # Apollo增强节点必填
DISCORD_BOT_TOKEN=      # Discord节点必填

所有密钥在启动时均为可选——仅当相关节点执行时，缺失的密钥才会触发运行时错误。

Start Development

启动开发环境

bash

undefined

bash

undefined

Terminal 1 — backend (PostgreSQL, Restate, all services)

终端1 — 后端服务（PostgreSQL、Restate及所有服务）

./dev.sh server

Terminal 2 — dashboard (SvelteKit at http://localhost:5173)

终端2 — 控制面板（SvelteKit，地址为http://localhost:5173）

./dev.sh dashboard

Or both at once

或者一次性启动所有服务

./dev.sh all

undefined

./dev.sh all

undefined

VS Code

Use the Dev Local All task to start server + dashboard in split terminals.

使用Dev Local All任务可在拆分终端中同时启动服务器和控制面板。

Development Commands

开发命令

bash

./dev.sh server               # Start backend services
./dev.sh dashboard            # Start frontend
./dev.sh all                  # Start everything
./dev.sh extension            # Build browser extension

./cleanup.sh                  # Stop everything, wipe Restate + DB
./cleanup.sh --no-db          # Stop services, keep database
./cleanup.sh --services       # Stop services only
./cleanup.sh --db-destroy     # Remove PostgreSQL container entirely

cargo build                   # Build without running PostgreSQL (uses .sqlx snapshots)
cargo test                    # Test without running PostgreSQL

bash

./dev.sh server               # 启动后端服务
./dev.sh dashboard            # 启动前端
./dev.sh all                  # 启动所有服务
./dev.sh extension            # 构建浏览器扩展

./cleanup.sh                  # 停止所有服务，清空Restate与数据库
./cleanup.sh --no-db          # 停止服务，保留数据库
./cleanup.sh --services       # 仅停止服务
./cleanup.sh --db-destroy     # 彻底移除PostgreSQL容器

cargo build                   # 构建项目，无需运行PostgreSQL（使用.sqlx快照）
cargo test                    # 测试项目，无需运行PostgreSQL

Infrastructure Nodes (Kubernetes)

基础设施节点（Kubernetes）

Only needed if using nodes like Postgres Database that provision K8s resources:

bash

curl -Lo ./kind https://kind.sigs.k8s.io/dl/v0.31.0/kind-$(uname -s | tr '[:upper:]' '[:lower:]')-amd64
chmod +x ./kind && sudo mv ./kind /usr/local/bin/kind

INFRASTRUCTURE_TARGET=local ./dev.sh server

仅在使用Postgres Database等需要部署K8s资源的节点时需要：

bash

curl -Lo ./kind https://kind.sigs.k8s.io/dl/v0.31.0/kind-$(uname -s | tr '[:upper:]' '[:lower:]')-amd64
chmod +x ./kind && sudo mv ./kind /usr/local/bin/kind

INFRASTRUCTURE_TARGET=local ./dev.sh server

The Weft Language

Weft语言

Core Concepts

核心概念

Nodes — typed computational units (LLM, HTTP, Human Query, Gate, etc.)
Connections — typed edges between node ports; compiler validates all types
Groups — collapse any set of nodes into a single reusable node
Durable execution — programs checkpoint via Restate; long pauses are transparent

节点 — 类型化计算单元（LLM、HTTP、人工查询、网关等）
连接 — 节点端口间的类型化连接；编译器会验证所有类型
组 — 将任意节点集合并为单个可复用节点
持久化执行 — 程序通过Restate创建检查点；长时间暂停对程序透明

Node Syntax

节点语法

weft

node_name = NodeType -> (output_port: OutputType) {
  label: "Human-readable name"
  config_key: "value"
}
node_name.input_port = other_node.output_port

weft

node_name = NodeType -> (output_port: OutputType) {
  label: "人类可读名称"
  config_key: "value"
}
node_name.input_port = other_node.output_port

Simple Example — Poem Generator

简单示例 — 诗歌生成器

weft

undefined

weft

undefined

Project: Poem Generator

项目：诗歌生成器

Description: Writes a short poem about any topic

描述：针对任意主题创作短诗

topic = Text { label: "Topic" value: "the silence between stars" }

llm_config = LlmConfig { label: "Config" model: "anthropic/claude-sonnet-4.6" systemPrompt: "Write a short, beautiful poem (4-6 lines) about the given topic." temperature: "0.8" }

poet = LlmInference -> (response: String) { label: "Poet" } poet.prompt = topic.value poet.config = llm_config.config

output = Debug { label: "Poem" } output.data = poet.response

---

topic = Text { label: "主题" value: "星辰间的寂静" }

llm_config = LlmConfig { label: "配置" model: "anthropic/claude-sonnet-4.6" systemPrompt: "针对给定主题创作一首简短优美的诗歌（4-6行）。" temperature: "0.8" }

poet = LlmInference -> (response: String) { label: "诗人" } poet.prompt = topic.value poet.config = llm_config.config

output = Debug { label: "诗歌" } output.data = poet.response

---

Built-in Node Catalog

内置节点目录

AI Nodes

AI节点

Node	Purpose
`LlmConfig`	Configure model, system prompt, temperature
`LlmInference`	Call an LLM, returns `response: String`

节点	用途
`LlmConfig`	配置模型、系统提示词、温度参数
`LlmInference`	调用LLM，返回 `response: String`

Data Nodes

数据节点

Node	Purpose
`Text`	Static or dynamic text value
`Number`	Numeric value
`Dict`	Key-value map
`List`	Ordered list
`Pack` / `Unpack`	Bundle/unbundle multiple values

节点	用途
`Text`	静态或动态文本值
`Number`	数值
`Dict`	键值对映射
`List`	有序列表
`Pack` / `Unpack`	打包/解包多个值

Flow Nodes

流程节点

Node	Purpose
`Gate`	Conditional branching
`HumanQuery`	Pause execution, send form to human, resume on response
`HumanTrigger`	Start a program from a human action

节点	用途
`Gate`	条件分支
`HumanQuery`	暂停执行，向人类发送表单，收到响应后恢复
`HumanTrigger`	通过人类操作启动程序

Communication Nodes

通信节点

Discord

Slack

Telegram

WhatsApp

Email

Discord

Slack

Telegram

WhatsApp

Email

Storage Nodes

存储节点

Postgres

Memory

Postgres

Memory

Enrichment Nodes

增强节点

Apollo

WebSearch

SpeechToText

Apollo

WebSearch

SpeechToText

Trigger Nodes

触发节点

Cron

, webhooks, polling

Cron

、Webhook、轮询

Utility Nodes

工具节点

Debug

Template

HTTP

Code

(Python execution)

Debug

Template

HTTP

Code

（Python执行）

Common Patterns

常见模式

Pattern 1 — LLM with Structured Config

模式1 — 带结构化配置的LLM

weft

undefined

weft

undefined

Project: Content Summarizer

项目：内容摘要器

Description: Summarizes a webpage given a URL

描述：针对给定URL的网页内容生成摘要

url_input = Text { label: "URL" value: "https://example.com/article" }

search = WebSearch -> (results: String) { label: "Fetch Content" } search.query = url_input.value

summarizer_config = LlmConfig { label: "Summarizer Config" model: "anthropic/claude-sonnet-4.6" systemPrompt: "Summarize the following content in 3 bullet points." temperature: "0.3" }

summarizer = LlmInference -> (response: String) { label: "Summarizer" } summarizer.prompt = search.results summarizer.config = summarizer_config.config

output = Debug { label: "Summary" } output.data = summarizer.response

undefined

url_input = Text { label: "URL" value: "https://example.com/article" }

search = WebSearch -> (results: String) { label: "获取内容" } search.query = url_input.value

summarizer_config = LlmConfig { label: "摘要器配置" model: "anthropic/claude-sonnet-4.6" systemPrompt: "将以下内容总结为3个要点。" temperature: "0.3" }

summarizer = LlmInference -> (response: String) { label: "摘要器" } summarizer.prompt = search.results summarizer.config = summarizer_config.config

output = Debug { label: "摘要" } output.data = summarizer.response

undefined

Pattern 2 — Human-in-the-Loop Approval

模式2 — 人机交互审批流程

weft

undefined

weft

undefined

Project: Content Approval Pipeline

项目：内容审批流水线

Description: AI drafts content, human approves before publishing

描述：AI生成内容草稿，经人类审批后发布

draft_config = LlmConfig { label: "Drafter Config" model: "openai/gpt-4o" systemPrompt: "Write a Twitter thread about the given topic. Be engaging." temperature: "0.7" }

topic = Text { label: "Topic" value: "distributed systems" }

drafter = LlmInference -> (response: String) { label: "Content Drafter" } drafter.prompt = topic.value drafter.config = draft_config.config

draft_config = LlmConfig { label: "草稿生成配置" model: "openai/gpt-4o" systemPrompt: "针对给定主题创作Twitter推文，风格要吸引人。" temperature: "0.7" }

topic = Text { label: "主题" value: "分布式系统" }

drafter = LlmInference -> (response: String) { label: "内容草稿生成器" } drafter.prompt = topic.value drafter.config = draft_config.config

Pauses execution indefinitely until a human responds

无限期暂停执行，直到人类响应

approval = HumanQuery -> (approved: Boolean, feedback: String) { label: "Human Approval" question: "Do you approve this draft for publishing?" } approval.content = drafter.response

gate = Gate -> (passed: String) { label: "Approval Gate" } gate.condition = approval.approved gate.value = drafter.response

publisher = Discord { label: "Publish to Discord" channel: "announcements" } publisher.message = gate.passed

undefined

approval = HumanQuery -> (approved: Boolean, feedback: String) { label: "人工审批" question: "是否批准此草稿发布？" } approval.content = drafter.response

gate = Gate -> (passed: String) { label: "审批网关" } gate.condition = approval.approved gate.value = drafter.response

publisher = Discord { label: "发布至Discord" channel: "announcements" } publisher.message = gate.passed

undefined

Pattern 3 — Conditional Branching with Gate

模式3 — 基于网关的条件分支

weft

undefined

weft

undefined

Project: Sentiment Router

项目：情感路由

Description: Routes messages based on sentiment analysis

描述：根据情感分析结果路由消息

message = Text { label: "Input Message" value: "This product is absolutely terrible!" }

sentiment_config = LlmConfig { label: "Sentiment Config" model: "anthropic/claude-haiku-3.5" systemPrompt: "Classify sentiment as 'positive' or 'negative'. Respond with one word only." temperature: "0.0" }

classifier = LlmInference -> (response: String) { label: "Sentiment Classifier" } classifier.prompt = message.value classifier.config = sentiment_config.config

is_negative = Gate -> (passed: String) { label: "Is Negative?" } is_negative.condition = classifier.response is_negative.value = message.value

alert = Slack { label: "Alert Team" channel: "customer-issues" } alert.message = is_negative.passed

undefined

message = Text { label: "输入消息" value: "这个产品简直糟透了！" }

sentiment_config = LlmConfig { label: "情感配置" model: "anthropic/claude-haiku-3.5" systemPrompt: "将情感分类为'positive'或'negative'。仅用一个单词回复。" temperature: "0.0" }

classifier = LlmInference -> (response: String) { label: "情感分类器" } classifier.prompt = message.value classifier.config = sentiment_config.config

is_negative = Gate -> (passed: String) { label: "是否负面？" } is_negative.condition = classifier.response is_negative.value = message.value

alert = Slack { label: "通知团队" channel: "customer-issues" } alert.message = is_negative.passed

undefined

Pattern 4 — Cron-Triggered Pipeline

模式4 — Cron触发的流水线

weft

undefined

weft

undefined

Project: Daily Digest

项目：每日摘要

Description: Sends a daily news digest every morning

描述：每天早上发送AI与技术新闻摘要

schedule = Cron { label: "Daily Trigger" expression: "0 8 * * *" }

news = WebSearch -> (results: String) { label: "Fetch News" } news.query = "AI and technology news today"

digest_config = LlmConfig { label: "Digest Config" model: "openai/gpt-4o-mini" systemPrompt: "Summarize these news items into a concise morning digest." temperature: "0.4" }

digest = LlmInference -> (response: String) { label: "Digest Writer" } digest.prompt = news.results digest.config = digest_config.config

send = Email { label: "Send Digest" to: "team@example.com" subject: "Your Daily AI Digest" } send.body = digest.response

undefined

schedule = Cron { label: "每日触发器" expression: "0 8 * * *" }

news = WebSearch -> (results: String) { label: "获取新闻" } news.query = "今日AI与技术新闻"

digest_config = LlmConfig { label: "摘要配置" model: "openai/gpt-4o-mini" systemPrompt: "将这些新闻条目总结为简洁的晨间摘要。" temperature: "0.4" }

digest = LlmInference -> (response: String) { label: "摘要生成器" } digest.prompt = news.results digest.config = digest_config.config

send = Email { label: "发送摘要" to: "team@example.com" subject: "你的每日AI摘要" } send.body = digest.response

undefined

Pattern 5 — Multi-Step Research Agent

模式5 — 多步骤研究Agent

weft

undefined

weft

undefined

Project: Research Agent

项目：研究Agent

Description: Researches a topic and produces a structured report

描述：研究指定主题并生成结构化报告

query = Text { label: "Research Query" value: "latest advances in protein folding" }

search = WebSearch -> (results: String) { label: "Search" } search.query = query.value

query = Text { label: "研究查询" value: "蛋白质折叠的最新进展" }

search = WebSearch -> (results: String) { label: "搜索" } search.query = query.value

Enrich with professional data

用专业数据增强内容

enrichment = Apollo -> (data: String) { label: "Enrichment" }

analyst_config = LlmConfig { label: "Analyst Config" model: "anthropic/claude-sonnet-4.6" systemPrompt: "You are a research analyst. Given search results, produce a structured report with: Executive Summary, Key Findings, Implications, and Further Reading." temperature: "0.2" }

pack_inputs = Pack -> (bundle: Dict) { label: "Combine Sources" } pack_inputs.search_results = search.results

analyst = LlmInference -> (response: String) { label: "Research Analyst" } analyst.prompt = pack_inputs.bundle analyst.config = analyst_config.config

store = Postgres { label: "Store Report" table: "research_reports" } store.data = analyst.response

notify = Slack { label: "Notify Team" channel: "research" } notify.message = analyst.response

---

enrichment = Apollo -> (data: String) { label: "数据增强" }

analyst_config = LlmConfig { label: "分析师配置" model: "anthropic/claude-sonnet-4.6" systemPrompt: "你是一名研究分析师。根据搜索结果，生成包含以下部分的结构化报告：执行摘要、关键发现、影响分析、进一步阅读。" temperature: "0.2" }

pack_inputs = Pack -> (bundle: Dict) { label: "合并数据源" } pack_inputs.search_results = search.results

analyst = LlmInference -> (response: String) { label: "研究分析师" } analyst.prompt = pack_inputs.bundle analyst.config = analyst_config.config

store = Postgres { label: "存储报告" table: "research_reports" } store.data = analyst.response

notify = Slack { label: "通知团队" channel: "research" } notify.message = analyst.response

---

Adding a Custom Node

添加自定义节点

New nodes are two files in the

catalog/

directory. No registration needed — the

inventory

crate auto-discovers nodes at startup.

新节点需在

catalog/

目录下创建两个文件，无需注册——

inventory

crate会在启动时自动发现节点。

Step 1 — Create the folder

步骤1 — 创建文件夹

catalog/
└── my_category/
    └── my_node/
        ├── backend.rs
        └── frontend.ts

catalog/
└── my_category/
    └── my_node/
        ├── backend.rs
        └── frontend.ts

Step 2 — Implement

backend.rs

步骤2 — 实现

backend.rs

rust

use weft_nodes::prelude::*;
use serde::{Deserialize, Serialize};

#[derive(Debug, Serialize, Deserialize)]
pub struct MyNodeConfig {
    pub label: String,
    pub my_setting: String,
}

#[derive(Debug, Serialize, Deserialize)]
pub struct MyNodeInputs {
    pub text: String,
}

#[derive(Debug, Serialize, Deserialize)]
pub struct MyNodeOutputs {
    pub result: String,
}

pub struct MyNode;

#[async_trait]
impl Node for MyNode {
    type Config = MyNodeConfig;
    type Inputs = MyNodeInputs;
    type Outputs = MyNodeOutputs;

    async fn run(
        config: Self::Config,
        inputs: Self::Inputs,
    ) -> Result<Self::Outputs, NodeError> {
        let result = format!("{}: {}", config.my_setting, inputs.text);
        Ok(MyNodeOutputs { result })
    }
}

rust

use weft_nodes::prelude::*;
use serde::{Deserialize, Serialize};

#[derive(Debug, Serialize, Deserialize)]
pub struct MyNodeConfig {
    pub label: String,
    pub my_setting: String,
}

#[derive(Debug, Serialize, Deserialize)]
pub struct MyNodeInputs {
    pub text: String,
}

#[derive(Debug, Serialize, Deserialize)]
pub struct MyNodeOutputs {
    pub result: String,
}

pub struct MyNode;

#[async_trait]
impl Node for MyNode {
    type Config = MyNodeConfig;
    type Inputs = MyNodeInputs;
    type Outputs = MyNodeOutputs;

    async fn run(
        config: Self::Config,
        inputs: Self::Inputs,
    ) -> Result<Self::Outputs, NodeError> {
        let result = format!("{}: {}", config.my_setting, inputs.text);
        Ok(MyNodeOutputs { result })
    }
}

Step 3 — Define

frontend.ts

步骤3 — 定义

frontend.ts

typescript

import type { NodeDefinition } from "@/types/nodes";

export const MyNode: NodeDefinition = {
  type: "MyNode",
  label: "My Node",
  icon: "sparkles",       // Lucide icon name
  category: "my_category",
  inputs: [
    { name: "text", type: "String", required: true },
  ],
  outputs: [
    { name: "result", type: "String" },
  ],
  config: [
    { name: "label", type: "string", default: "My Node" },
    { name: "my_setting", type: "string", default: "prefix" },
  ],
};

After adding both files, run

./dev.sh server

— the node is available immediately.

typescript

import type { NodeDefinition } from "@/types/nodes";

export const MyNode: NodeDefinition = {
  type: "MyNode",
  label: "My Node",
  icon: "sparkles",       // Lucide图标名称
  category: "my_category",
  inputs: [
    { name: "text", type: "String", required: true },
  ],
  outputs: [
    { name: "result", type: "String" },
  ],
  config: [
    { name: "label", type: "string", default: "My Node" },
    { name: "my_setting", type: "string", default: "prefix" },
  ],
};

添加两个文件后，运行

./dev.sh server

——节点会立即生效。

Type System

类型系统

Weft has generics, unions, type variables, and null propagation.

Type	Description
`String`	Text value
`Number`	Numeric value
`Boolean`	True/false
`Dict`	Key-value map
`List`	Ordered collection
`T?`	Nullable type — null propagates through the graph
`T \| U`	Union type

The compiler validates every connection before execution. Type mismatches, missing required connections, and broken architecture are caught at compile time.

Weft支持泛型、联合类型、类型变量和空值传播。

类型	描述
`String`	文本值
`Number`	数值
`Boolean`	布尔值（真/假）
`Dict`	键值对映射
`List`	有序集合
`T?`	可空类型——空值会在图中传播
`T \| U`	联合类型

编译器会在执行前验证所有连接。类型不匹配、缺失必要连接和架构问题都会在编译阶段被捕获。

Project Layout Reference

项目结构参考

weft/
├── catalog/                # Node definitions — source of truth
│   ├── ai/                 # LlmConfig, LlmInference
│   ├── code/               # Python execution
│   ├── communication/      # Discord, Slack, Telegram, WhatsApp, Email, X
│   ├── data/               # Text, Number, Dict, List, Pack, Unpack
│   ├── enrichment/         # Apollo, WebSearch, SpeechToText
│   ├── flow/               # Gate, HumanQuery, HumanTrigger
│   ├── storage/            # Postgres, Memory
│   └── triggers/           # Cron, webhooks, polling
├── crates/
│   ├── weft-core/          # Type system, compiler, executor, Restate objects
│   ├── weft-nodes/         # Node trait, registry, sandbox, node runner
│   ├── weft-api/           # REST API (triggers, files, infra, usage)
│   └── weft-orchestrator/  # Restate services + Axum project executor
├── dashboard/              # Web UI (SvelteKit + Svelte 5)
├── extension/              # Browser extension for human-in-the-loop (WXT)
└── scripts/
    └── catalog-link.sh     # Symlinks catalog into crates + dashboard

weft/
├── catalog/                # 节点定义——事实来源
│   ├── ai/                 # LlmConfig、LlmInference
│   ├── code/               # Python执行
│   ├── communication/      # Discord、Slack、Telegram、WhatsApp、Email、X
│   ├── data/               # Text、Number、Dict、List、Pack、Unpack
│   ├── enrichment/         # Apollo、WebSearch、SpeechToText
│   ├── flow/               # Gate、HumanQuery、HumanTrigger
│   ├── storage/            # Postgres、Memory
│   └── triggers/           # Cron、Webhook、轮询
├── crates/
│   ├── weft-core/          # 类型系统、编译器、执行器、Restate对象
│   ├── weft-nodes/         # 节点 trait、注册表、沙箱、节点运行器
│   ├── weft-api/           # REST API（触发器、文件、基础设施、使用统计）
│   └── weft-orchestrator/  # Restate服务 + Axum项目执行器
├── dashboard/              # Web UI（SvelteKit + Svelte 5）
├── extension/              # 人机交互浏览器扩展（WXT）
└── scripts/
    └── catalog-link.sh     # 将catalog链接到crates和dashboard

Troubleshooting

故障排除

./dev.sh server

fails immediately

./dev.sh server

立即失败

Ensure Docker is running (
```
docker ps
```
)
Check that ports 5432 (Postgres), 8080 (Restate), and 9070 (Restate admin) are free
Run
```
./cleanup.sh
```
then retry

确保Docker正在运行（执行
```
docker ps
```
检查）
检查端口5432（Postgres）、8080（Restate）和9070（Restate管理端）是否空闲
运行
```
./cleanup.sh
```
后重试

Node shows "API key missing" at runtime

节点在运行时显示“API密钥缺失”

Add the required key to
```
.env
```

Restart the server (

./cleanup.sh --services && ./dev.sh server

)

将所需密钥添加到
```
.env
```
文件中

重启服务器（执行

./cleanup.sh --services && ./dev.sh server

）

Type mismatch compiler error

编译器报错类型不匹配

Check that the output port type of the source node matches the input port type of the destination node
Use
```
Pack
```
to bundle multiple values into a
```
Dict
```
before passing to a node that expects
```
Dict
```
Nullable types (
```
T?
```
) can propagate null — use a
```
Gate
```
node to guard against null before consuming

检查源节点的输出端口类型是否与目标节点的输入端口类型匹配
使用
```
Pack
```
将多个值打包为
```
Dict
```
后，再传递给需要
```
Dict
```
类型的节点
可空类型（
```
T?
```
）会传播空值——在使用前用
```
Gate
```
节点防范空值

Restate state is stale after a crash

崩溃后Restate状态过期

bash

./cleanup.sh        # Wipes Restate journal and DB, full reset

bash

./cleanup.sh        # 清空Restate日志和数据库，完全重置

cargo build

fails without running Postgres

未运行Postgres时

cargo build

失败

The

.sqlx

directory is committed — offline mode is supported:

bash

cargo build         # Works without a running database
cargo test          # Works without a running database

.sqlx

目录已提交——支持离线模式：

bash

cargo build         # 无需运行数据库即可构建
cargo test          # 无需运行数据库即可测试

New node not appearing in dashboard

新节点未在控制面板中显示

Confirm both

backend.rs

and

frontend.ts

exist under

catalog/<category>/<node>/

Run
```
./scripts/catalog-link.sh
```
manually then restart the server
Check server logs for inventory discovery errors

确认

backend.rs

和

frontend.ts

都存在于

catalog/<category>/<node>/

目录下

手动运行
```
./scripts/catalog-link.sh
```
后重启服务器
检查服务器日志中的节点发现错误

Key Resources

关键资源

Documentation: https://weavemind.ai/docs
Getting Started: https://weavemind.ai/docs/hello-world
Language Reference: Nodes, Connections, Types, Groups, Parallel — all at https://weavemind.ai/docs
Architecture:
```
DESIGN.md
```
in the repo root
Roadmap:
```
ROADMAP.md
```
in the repo root
Contributing:
```
CONTRIBUTING.md
```
in the repo root
Community Discord: https://discord.com/invite/FGwNu6mDkU
License: O'Saasy License (MIT + no competing hosted service) — https://osaasy.dev

文档： https://weavemind.ai/docs
快速入门： https://weavemind.ai/docs/hello-world
语言参考： 节点、连接、类型、组、并行等内容均在https://weavemind.ai/docs
架构： 仓库根目录下的
```
DESIGN.md
```
路线图： 仓库根目录下的
```
ROADMAP.md
```
贡献指南： 仓库根目录下的
```
CONTRIBUTING.md
```
社区Discord： https://discord.com/invite/FGwNu6mDkU
许可证： O'Saasy许可证（MIT + 禁止竞争托管服务）—— https://osaasy.dev

weft-ai-language

Original

Translation

Weft AI Language

Weft AI 语言

Installation & Setup

安装与配置

Prerequisites

前置要求

Clone and Configure

克隆与配置

Edit .env — add your API keys

编辑.env — 添加你的API密钥

Environment Variables (.env)

环境变量（.env）

Start Development

启动开发环境

Terminal 1 — backend (PostgreSQL, Restate, all services)

终端1 — 后端服务（PostgreSQL、Restate及所有服务）

Terminal 2 — dashboard (SvelteKit at http://localhost:5173)

终端2 — 控制面板（SvelteKit，地址为http://localhost:5173）

Or both at once

或者一次性启动所有服务

VS Code

VS Code

Development Commands

开发命令

Infrastructure Nodes (Kubernetes)

基础设施节点（Kubernetes）

The Weft Language

Weft语言

Core Concepts

核心概念

Node Syntax

节点语法

Simple Example — Poem Generator

简单示例 — 诗歌生成器

Project: Poem Generator

项目：诗歌生成器

Description: Writes a short poem about any topic

描述：针对任意主题创作短诗

Built-in Node Catalog

内置节点目录

AI Nodes

AI节点

Data Nodes

数据节点

Flow Nodes

流程节点

Communication Nodes

通信节点

Storage Nodes

存储节点

Enrichment Nodes

增强节点

Trigger Nodes

触发节点

Utility Nodes

工具节点

Common Patterns

常见模式

Pattern 1 — LLM with Structured Config

模式1 — 带结构化配置的LLM

Project: Content Summarizer

项目：内容摘要器

Description: Summarizes a webpage given a URL

描述：针对给定URL的网页内容生成摘要

Pattern 2 — Human-in-the-Loop Approval

模式2 — 人机交互审批流程

Project: Content Approval Pipeline

项目：内容审批流水线

Description: AI drafts content, human approves before publishing

描述：AI生成内容草稿，经人类审批后发布

Pauses execution indefinitely until a human responds

无限期暂停执行，直到人类响应

Pattern 3 — Conditional Branching with Gate

模式3 — 基于网关的条件分支

Project: Sentiment Router

项目：情感路由

Description: Routes messages based on sentiment analysis

Environment Variables (
`.env`
)

环境变量（
`.env`
）

Step 2 — Implement
`backend.rs`

步骤2 — 实现
`backend.rs`

Step 3 — Define
`frontend.ts`

步骤3 — 定义
`frontend.ts`

`./dev.sh server`
fails immediately

`./dev.sh server`
立即失败

`cargo build`
fails without running Postgres

未运行Postgres时
`cargo build`
失败