AI tools, coding

MCP toolkit for coding — semantic retrieval and editing capabilities, "the IDE for your agent".

Secure, fast, extensible sandbox runtime for executing AI-agent code.

LLM inference in C/C++ — runs models locally on CPU/GPU across platforms with minimal dependencies.

High-throughput, memory-efficient inference and serving engine for LLMs, built around PagedAttention.

Frontend stack and generative UI for in-app AI agents and copilots across React, Angular, and mobile.

Open-source LLM app development platform — agentic workflows, RAG pipelines, and a visual builder.

AI-driven development platform — autonomous coding agents that write code, fix bugs, and ship features.

LLM-friendly web crawler and scraper — turns any site into clean markdown for agents and RAG.

Typed Python framework for agents — every model, realtime voice, image generation, and embeddings.

Local UI to run, fine-tune, and train LLMs and diffusion models — supports GGUF, MLX, and reinforcement learning workflows.

AI-first code editor forked from VS Code, built around model-driven completions, chat, and agent workflows.

Glamourous agentic coding for all — terminal-native AI coding assistant from the Charm/Bubbletea team, written in Go.

Microsoft's R&D agent for automating data-driven AI research and development.

Most-adopted open-source AI coding extension for VS Code — bring-your-own-model agent with no markup, 5M+ installs.

All-in-one open-source agentic coding extension for VS Code — plan, build, and iterate with any model.

OpenAI's terminal-native coding agent — runs locally or as a cloud agent, bundled with ChatGPT Plus/Pro and the API.

Open-source IDE extension for building custom AI code assistants — model-agnostic autocomplete, chat, and agents.

StackBlitz's in-browser AI app builder — prompt to full-stack app with live preview and one-click deploy.

AI agent toolkit: coding agent CLI, unified LLM API, TUI & web UI libraries, Slack bot, vLLM pods.

Open-source extensible AI agent that goes beyond code suggestions — install, execute, edit, and test with any LLM.

Full-stack framework to build, run, and manage multi-agent systems and agent platforms at scale.

Alibaba's latest open-weight LLM series, strong on coding and reasoning with hybrid thinking modes.

Fast AI gateway — call 100+ LLM providers through one OpenAI-compatible API, as proxy or SDK.

High-performance serving framework for LLMs and vision-language models — RadixAttention and fast structured output.

Microsoft's official Playwright MCP server — structured accessibility-tree browser automation for AI coding agents.

The fast, Pythonic way to build MCP servers and clients — decorators over boilerplate, from the Prefect team.

Fullstack MCP framework — build MCP apps for ChatGPT and Claude, and MCP servers for AI agents.

Anthropic's official CLI for Claude — agentic coding in the terminal with file editing, shell, and IDE integrations.

Open-source AI pair-programming CLI that edits code in your local git repo across many models.

GitHub's AI pair programmer — inline code suggestions and chat across editors. Slug tracks the copilot.vim plugin used by data-collector.

Agentic IDE with Cascade flows, deep context understanding, and multi-file edits.

Google's agentic, Gemini-powered browser IDE with task-budgeted autonomous coding flows.

JetBrains' agentic coding assistant across every JetBrains IDE and CLI, with per-task credit usage.

Quality-first AI coding platform — code review, test generation, and agentic edits with a free tier.

Terminal-native open-source AI coding agent (by the SST team) — provider-agnostic, runs any model.

Browser-based agent that builds and deploys full apps from a prompt inside the Replit cloud IDE.

Alibaba's open-source terminal coding agent, tuned for the Qwen3-Coder models — provider-agnostic.

Fully local Manus AI alternative — no APIs, no monthly bills. Autonomous agent that thinks, browses the web, and codes.

Ghostty-based macOS terminal with vertical tabs and notifications for AI coding agents.

High-performance, multiplayer code editor from the creators of Atom and Tree-sitter, with a built-in AI agent.

Persistent memory across agent sessions — captures, compresses, and retrieves context automatically.

Coding-agent harness for large, complex codebases, tuned for token efficiency.

Open-source, Claude Code-compatible coding agent that runs anywhere against any model.

A whole team of AI dev agents inside VS Code (Cline fork).

Self-evolving agent that grows a skill tree from a ~3.3K-line seed with full system control.

Unified, efficient fine-tuning of 100+ LLMs and VLMs — LoRA/QLoRA, full-tune, and a no-code web UI.

Microsoft's programming framework for agentic AI — multi-agent conversations, tool use, and orchestration.

Framework for orchestrating role-playing, autonomous AI agents into collaborating crews.

1000+ tool integrations plus a sandboxed workbench and auth layer for AI agents.

Production-grade engineering skills for AI coding agents, curated by Addy Osmani.

Compresses tool outputs, logs, and RAG chunks before they reach the LLM — 20% fewer tokens for coding agents, same answers.

Visual framework for building AI agents and RAG pipelines — drag-and-drop flows, deployable as APIs.

Coding agent that runs LLMs locally on your machine — natural-language commands executed as code.

Self-hosted all-in-one AI assistant — chat, RAG, and agents over your own documents and models.

Open-source RAG engine with deep document understanding — truthful answers grounded in complex-format files.

Universal memory layer for AI agents — cross-session persistence and personalization via one API.

Low-code visual builder for AI agents and LLM apps — drag-and-drop nodes, deploy as APIs or chatbots.

OpenAI's lightweight framework for multi-agent workflows — handoffs, guardrails, and tracing in plain Python.

The runtime your coding agents live on — telemetry, sandboxing, and orchestration in one layer.

Agent harness optimization system for Claude Code, Codex, and OpenCode — skills, instincts, memory, security, and research-first development.

MCP server that feeds up-to-date, version-specific library documentation to LLMs and AI code editors.

Kanban board for orchestrating Claude Code, Codex, and other coding agents in parallel with review and merge in one place.

TypeScript framework for AI apps and agents — workflows, RAG, evals, memory, and a local dev playground.

Platform for stateful agents with long-term memory that learn and self-improve across sessions (formerly MemGPT).

Open-source AI orchestration framework for building context-engineered, production-ready LLM applications and RAG pipelines.

One local control plane for every AI agent — route across models, fuse capabilities, and orchestrate tools while staying in control.

The AI toolkit for TypeScript from the creators of Next.js — streaming UI, tool calling, and agents across providers.

Context window optimization for coding agents — sandboxes tool output, persists session memory, and enforces routing via MCP and hooks.

MCP server that feeds Figma layout data to AI coding agents like Cursor for design-to-code work.

Context-aware AI coding agent for large codebases — deep retrieval across the whole repo, IDE + CLI.

JavaScript in-page GUI agent from Alibaba — control web interfaces with natural language by dropping a script into any page.

Google's open-source MCP server for databases — exposes SQL and NoSQL sources as tools for AI agents.

AWS's agentic coding assistant — code generation, AWS-aware suggestions, security scans, and CLI agent.

Vercel's generative UI tool — prompt to React/Tailwind components and full-stack apps, deploy on Vercel.

Terminal coding agent — hash-anchored edits, optimized tool harness, LSP, Python, browser, subagents.

Gives AI agents read access across Twitter, Reddit, YouTube, GitHub, Bilibili, and Xiaohongshu.

Self-healing browser harness that lets LLMs complete web tasks reliably.

GEP-powered self-evolving engine with auditable evolution via Genes, Capsules, and Events.

Stealth headless browser for AI agents that bypasses Cloudflare and bot detection.

Cross-platform desktop assistant for Claude Code, Codex, and OpenCode — provider switching, MCP management, and prompt presets.

Privacy-focused AI coding assistant with on-prem/air-gapped options and per-team model personalization.

AI app builder that turns prompts into full-stack web apps with Supabase backends and live editing.

Alibaba's Qwen-powered AI coding assistant — completions, chat, and agentic tasks across IDEs, popular in China.

Open-source infrastructure for Computer-Use Agents — sandboxes, SDKs, and benchmarks for desktop-controlling AI.

Identity, communication, and payments layer for AI agents.

Open-source autonomous general agent — a no-invite-code alternative to Manus.

Agentic skills framework and software development methodology for AI coding agents.

Makes AI coding agents think like lazy senior developers — ship the minimum code that works.

Agent engineering framework — chains, tools, and memory for building context-aware LLM applications.

Open-source community-driven agent harness — extensible coding agents shaped by contributor plugins.

Microsoft's official inference framework for 1-bit LLMs — runs large models efficiently on a single CPU.

GitHub's official MCP server — repos, issues, pull requests, and Actions exposed as tools for AI agents.

Microsoft's SDK for integrating LLMs into apps with plugins, planners, and agents — Python, C#, and Java.

Takes a GitHub issue and tries to fix it automatically with the LM of your choice — the NeurIPS 2024 autonomous software engineering agent.

MCP server that lets AI agents run 150+ offensive and defensive security tools for automated pentesting and security research.

An open-source AI agent that brings the power of Gemini directly into your terminal.

Gives AI agents good taste — steers generated UIs and copy away from boring generic slop.

Local-first desktop workspace for Claude Code and agents — multi-agent Git workflows on your machine.

Multi-agent framework framed as the first AI software company — assigns roles like PM, architect, and engineer toward natural language programming.

ByteDance's open-source SuperAgent harness — researches, codes, and creates with sandboxes, memory, tools, and subagents.

Agent meta-harness for Claude — multi-agent swarms, adaptive memory, self-learning, RAG, native Claude Code/Codex integration.

Lightweight open-source AI agent for your tools, chats, and workflows.

DeepSeek-native AI coding agent for your terminal — engineered around prefix-cache stability.

High-performance code intelligence MCP server that indexes codebases into a persistent knowledge graph — 158 languages, sub-ms queries, single static binary.