AI Agent 工具
從能替你執行任務的 Agent,到用來建立多步驟工作流的框架,集中查看能力邊界、適用情境與來源資訊。
查看 AI Agent 工具推薦 →我們如何劃分 Agent 工具
- 是否能執行多步驟任務,而不只是產生一次性回答。
- 是否支援工具呼叫、檔案操作、瀏覽器或外部服務連線。
- 是否提供記憶、狀態管理、人工確認或失敗復原能力。
- 是否清楚說明模型支援、部署邊界、權限與資料處理方式。
按使用場景瀏覽
已發布的 Agent 工具
依品質訊號與資訊完整度排序,先看目錄,再開啟單一工具查看來源與限制。
Cossistant is the open-source AI and human support framework for React and Next.js apps, with programmable actions, custom UI, and code-first workflows.
Dash is the open-source Python framework for interactive data apps and dashboards. No HTML, CSS, or JavaScript required. Get started free.
Connect GitHub, get branch previews and production MCP endpoints with managed TLS, env vars, and deploy history. Promote, roll back, and wire CI, then pair w...
Build an AI contract review tool that identifies risks, key clauses, and obligations for legal professionals. Start building with Modelence, a full-stack AI app builder with authentication, database, and deployment built in.
the end-to-end failure layer for teams building ai agents. observe every run, find where agents fail, write a policy that stops it, and enforce it in realtime, across claude code, codex, cursor, gemini, copilot and more. free: npm i -g failproofai
Let your users generate, review, and refine production-ready interfaces inside your application with Puck.
OpenTelemetry-native tracing with runtime policy enforcement, prompt registry & evals, cost attribution, and EU AI Act evidence. One SDK for LangChain, CrewAI, OpenAI Agents SDK & more.
Revolutionising AI Accessibility Praison AI simplifies the development of AI-powered multi-agent systems, making complex AI capabilities accessible to businesses of all sizes. Praison AI Customer Service Real-Time Customer InteractionEnable responsive customer support with voice-enabled AI, improving call resolution and customer satisfaction through real-time responses. AI Personalisation Engage Customers with Tailored ExperiencesPraison AI’s personalisation services deliver […]
ConnectOnion is an open-source Python toolkit for AI agents. `co create` scaffolds a project that already runs — shell, file editing, search and a model wired in — and you edit it from there. Supports OpenAI, Anthropic, and Gemini.
Agent Kernel 是一個開源的企業級 AI 代理作業系統,支援建置、測試和部署可擴展且合規的 AI 代理。它與 OpenAI、LangGraph、CrewAI、Google ADK、Smolagents 和 Pydantic AI 等框架無縫整合,並可在 AWS、Azure、GCP 或 Kubernetes 上執行。內建訊息傳遞、記憶、知識庫、護欄、沙箱程式碼執行和可觀測性功能。
TypeScript-native multi-agent framework: one goal in, a self-organizing team of AI agents runs it in parallel — in your environment, cloud or fully offline.
OpenAgent 是一個開源的個人AI助手平台,支援連接30多個模型提供商,透過MCP工具使用建構智慧代理,並部署到20多個訊息管道。
Symbiont 是一個 Rust 原生的執行環境,用於在明確的策略、身份和稽核控制下執行 AI 代理和工具。它專注於在真實環境中運行代理,處理不可信工具、敏感資料、核准邊界和稽核要求。
OpenLIT is an open-source AI Engineering platform for LLM observability, evaluations, prompt management, and cost tracking - built on OpenTelemetry, fully self-hostable.
A typed Rust runtime for model calls, tools, memory, multi-agent workflows, realtime voice, and open agent protocols.
Get started How to get things done with AgentQL Check out our extensive guides to learn how to use AgentQL to automate web interactions and extract data from w...
Build AI support agents with human escalation, full context handoff, and production-ready chat UI in Angular.
Intercom is the only helpdesk with a natively integrated AI Agent, Fin, meaning every customer conversation improves the next one, enabling perfect experiences that were never possible before.
Give it a direction - aeon leverages 78 skills to get it done. No approval loops. No babysitting. Configure once, forget forever.
AI 视频会被限流吗?按规则标识的 AIGC 内容属有条件允许,刻意去标识才是高风险。本文讲清国标 GB 45438-2025 显式与隐式标识要求、抖音测评形态与数字人五类目规则,以及 ClipForge 合规默认开:自动写入隐式元数据标识、默认烧录「AI 生成」角标,限流自检与违禁词扫描逐项给改法。
Network-AI is a TypeScript/Node.js multi-agent orchestration layer with shared-state locking, permission gating, audit trails, FSM governance, and 32 framework adapters.
OpenAI Agents SDK 是一個用於構建現代 AI 代理的框架,支援上下文管理、工具呼叫、多代理協作等功能。它提供了沙盒環境,使代理能夠在真實檔案系統上操作,執行命令、編輯檔案和處理大型文件集。
Pest is the elegant PHP testing framework built for developers and AI agents — with browser testing, AI evals, an agent browser that verifies your coding agent's work end-to-end, and the Tia Engine for near-instant test runs.
Register built-in or custom tools and control parallel vs sequential execution per turn
Cursor plans, writes, and reviews code using AI agents that understand your entire codebase. Everywhere you work: Desktop, CLI, Web & Mobile.
Spring AI Alibaba 是一個面向 Java 開發者的 Agentic AI 框架,基於 DAG 圖核心概念,可輕鬆實現單智慧體、多智慧體和複雜工作流編排。
It’s hard to move fast if you’ve got a clunky and disorganized workspace. Centralize all your knowledge in Notion instead.
Strikethroo 是一個基於 Agent Skills 的 AI 任務管理工具,透過兩個命令(st-create-plan 和 st-execute-blueprint)將任務分解為計畫並執行,支援可選的自動化程式碼審查。
Understand the differences between LangChain, LangGraph, and Deep Agents and when to use each one
For all the agents you use today, and the ones you'll use tomorrow.
Run open and frontier LLMs on your voice data—fast open models for the cleanup pass, frontier models for the reasoning pass, one API key for both.
AXIS is open source tooling and a scoring framework to measure how well services work for AI agents. Run real scenarios, get a 0–100 score, and fix what's slowing agents down.
boringSQL 提供工具集,幫助開發者測試和優化 PostgreSQL 查詢,其中 RegreSQL 是用於 SQL 查詢回歸測試的開源工具。
A private, self-hosted AI agent: one brain — a single fast Go binary — with eight legs, like an octopus, reaching your terminal, VS Code, Obsidian, SDK, web, desktop, IM, and mobile. Any model, zero telemetry; your data never leaves your machine.
Read ebooks and take online courses while the author is still writing them — and get every update free. Or write, publish, and sell your own in-progress book or course on Leanpub.
Winston Koh 是一位駐新加坡的 AI 系統架構師,專注於構建自主智慧體、可靠的業務工作流和生產級軟體。他提供自由職業和全職服務,包括 AI 系統架構、本地優先 AI 架構(Athena)以及面向中小企業的網站開發。
One YAML = Any Agent. A Rust framework for building AI agents from a single YAML specification.
FAQ
AI Agent 工具常見問題
AI Agent 和普通聊天工具有什麼差異?
Agent 通常可以規劃並執行多個步驟,呼叫工具或操作外部系統;普通聊天工具更側重對話生成內容。
應該選 Agent 產品還是 Agent 框架?
需要立即完成任務時先看產品;需要把流程嵌入自己的應用並控制模型與權限時,優先比較框架或開發工具。
Mengbi 如何確認分類?
分類依據公開來源中的功能、文件與產品定位,並結合自動品質檢查;頁面會保留來源線索與最近核驗資訊。