Openai agents

A coding agent in 5 files, plus a skeptic that catches it faking the fix — verify AI code with a check the agent never saw. Runs on any model (OpenRouter/OpenAI/Ollama).

欲しい機能を言うと、アプリが自分で作る。ただし暴走はしない。安全な自己進化型デスクトップAIエージェント / Self-evolving desktop AI agent with safety gates

A persistent cognitive entity — one agent, one memory, one timeline across all platforms. Self-editing knowledge graph, cross-platform identity, P2P network.

Async FastAPI LLM agent with LangGraph routing, scope-aware tools, deterministic verification, bounded recovery, evals, and durable traces.

Drop-in compression for LLM agent tool outputs: cut ~99% of tool-result tokens while keeping (often improving) answer accuracy. Provider-agnostic, with a faithfulness benchmark and an OpenAI-compatible proxy.

Agent-ready project documentation template with START/STATE/TODO/DONE, workflows, lightweight tooling, and LLM-friendly project memory. Independent project - not affiliated with Anthropic, OpenAI or Google.

Local AI Linux troubleshooting agent with llama.cpp, safe command execution, system diagnostics, Docker support, CLI, and web GUI.

MCP server for Cognigy.AI - 132 tools that let Claude, Cursor & other AI assistants build, configure, test & operate conversational AI agents via the Model Context Protocol.

AI Research Agent 2026 🔍 - Autonomous Deep Web Explorer & Adversarial Synthesis

Autonomous LLM agent auditing any CSV for statistical and intersectional bias. Chi-squared, Cramer's V, LLM-as-judge. Live demo on HF Spaces.

Self-hosted agent infrastructure framework — orchestration, hybrid RAG, workflows, tools and full-trace audit behind one API. Ships with a React console, SSE streaming, and support for any OpenAI-compatible LLM.

Build a production-grade multi-agent AI framework for collaborative software development with clean orchestration and repeatable workflows

Async multi-agent AI library for Python — self-improving agent loops, APEX synthesis, bounded agent spawning, multimodal input, and true token streaming across 100+ LLM providers.

TypeScript framework for orchestrating collaborative AI agents with task scheduling, shared memory, tools, and multi-provider LLM support.

C++23 framework for building LLM-powered agents with tool use, structured output, multi-provider

Typed, model-agnostic LLM agents in Swift — a pydantic-ai port built on Apple FoundationModels. Anthropic Claude, OpenAI, Google Gemini, and on-device, with tools, structured output, streaming, and a graph-based run engine.

Autonomous multi-agent research: plan → research → synthesize → critique → finalize. Live agent graph, HITL approval, RAG, Reflexion, streaming SSE, eval gate. Next.js 16, Drizzle, Postgres. Groq llama-3.3-70b + Tavily when keys are set; otherwise a deterministic grounded simulator with the same UI. Free-tier Vercel demo. OpenAI-compatible keys.

AI-powered agent that autonomously discovers and enters free sweepstakes using browser-use

TypeScript SDK for orchestrating Agentic Systems — concurrency, structured output, cost control, and consensus as first-class primitives.

Build an AI chatbot with LangGraph, RAG, and Gemini for accurate answers, workflow automation, and easy Streamlit access

TalentFlow HR — modern dark-themed HR dashboard for IT staffing & consulting firms. Drag-and-drop candidate pipeline, job requisitions, interview tracker, analytics, and a built-in AI assistant (Ollama / OpenAI / Anthropic / Gemini) with offline template fallback. React + Vite + Tailwind. Zero backend, zero cost.

Open-source proactive on-screen hints for Windows — an MIT-licensed, bring-your-own-VLM alternative to closed 'magic pointer'-style assistants. Backends: Anthropic Claude, OpenAI, OpenRouter, Gemini, Ollama, LM Studio, and any OpenAI-compatible endpoint.

Native AI assistant for SOLIDWORKS. In-process COM add-in, agentic loop, governed write tools. MIT licensed. v0.1.1 public beta; v1.0 pending live R5 validation.

墨灵写作 — AI 驱动的网文创作平台 | AI-Powered Novel Writing Platform

Production QuickBooks MCP Server: AI-powered assistant, Azure/Docker deployment, OAuth automation, unified API. Fork of Intuit's official implementation with enhanced features: customer search, transaction analytics, multi-provider AI (GPT/Claude), and enterprise-ready deployment.

Browse a curated list of top AI tools, models, and frameworks to accelerate your development work.

Voice-driven extended reality. GPT Realtime hears you, GPT-5.6 Sol plans agentic, and Reframe segments the object and rebuilds the background from monocular depth and TSDF fusion without LiDAR.

AI-powered side panel for DHIS2 (Chrome MV3). 32 tools — analytics, tracker writes, atomic metadata authoring (programs, datasets, category combos, sharing), rule/indicator audit + repair, backups. Any OpenAI-compatible provider; defaults to local Ollama, no API key. Authenticates via active DHIS2 tab session.

🦈 Terminal-only AI assistant with an animated neon voice orb, hands-free voice chat, neural TTS, and a safe agent toolset. Providers: DeepSeek, OpenAI, Claude, Gemini, Ollama.

Complete, verified command reference and comparison guide for Claude Code (Anthropic) and Codex CLI (OpenAI): every slash command, flag, and shortcut. FR + EN PDFs.

MCP server: token cost math for LLM API calls, 69 models across 17 providers, prices verified by ComparEdge

A tested library of 25+ ready-made tools for LLM agents — web, files, data, math, text, datetime — with OpenAI schemas, Pydantic validation and pytest coverage

Universal LLM provider compatibility layer — normalizes provider-specific payload differences (thinking formats, output budgets, tool pairing, media transport) into a single first-match-wins dispatch pipeline. Plus dynamic version detection with strategy-pattern version parsing.

🏓 Bounce — Multi-layer AI provider failover gateway. UPS for your LLM tools. 多级 AI 提供商故障转移网关，让你的爱AI永不停机。

🛰️ Talk to any AI agent runtime through one TypeScript client. Provider-agnostic, ESM, zero-dependency.

Multi-agent SQL assistant that turns natural-language questions into safe, human-approved SQL queries - connects to Postgres, MySQL, SQL Server, or any spreadsheet, with live query streaming and auto-generated charts.

Universal engineering kernel and reusable skill for agent-led repositories

JARVIS, but real. A local-first AI assistant for Windows that sees your screen, controls your apps via UI automation, speaks back with neural TTS, and runs on Ollama — with hot-swappable cloud fallback to OpenAI/Anthropic/Gemini. Includes a React dashboard with real-time SSE streaming.

A personal assistant that runs on your own machine, does real work on it, and writes itself new abilities when the ones it has are not enough.

The fullstack framework for building AI Agents in Ruby on Rails. Build, test, deploy, and monitor in one platform. Advanced alternative to RubyLLM and Active Agent.

A voice-first AI personal finance assistant with secure voice authentication that enables hands-free expense and income logging, budget management, and natural language financial queries using FastAPI, React, OpenAI Whisper, and SQLite.

🤖 Sofia IA - Assistente Virtual Inteligente para WhatsApp com Google Calendar e CRM

Design, audit, and test reliable routing contracts for Codex skills.

AI Hedge Fund Framework 2026 – Multi-agent AI trading simulation with 10+ investor personas. Educational backtesting & paper trading. Setup.exe included.

OpenAI Codex Skill and system prompt generator for AI agents: build six-part XML LLM prompts, then execute them immediately.

Open-source MCP vision server that gives text-only LLMs and AI agents image understanding, OCR, visual analysis, UI inspection, and multimodal capabilities.

Open-source AI agent framework for Flutter. Build persistent mobile agents with 1000+ app integrations via Composio. MIT licensed, 3-min setup.

A lightweight Python framework for building OpenAI Agents with pluggable model providers.

Local-first context layer for AI coding — index repos, cut token waste via MCP + Desktop

BridgeNode agent skill and buyer-side examples — AI inference via x402 on Solana. No API keys, pay per request with USDC.

AI-powered developer assistant extension for Code Editors with scalable architecture, multi-model support, and future multi-IDE compatibility.

Spec-first, embeddable agent harness runtime - a control plane for loop-engineered AI agents. Governed tool calls through one gateway, one normalized event stream, durable sessions.

🗣️ AgentRound — 多模型 AI 圆桌讨论框架。让多个 AI 模型围绕同一主题展开多轮辩论，支持并发调用、人类实时介入、智能 Token 裁剪、Rich 终端渲染和 Markdown 导出。兼容任意 OpenAI 兼容 API。

Governed, auditable AI review platform: the review and approval work a team does by hand, as a webhook-triggered LangGraph agent with human-in-the-loop approval, Langfuse tracing, an eval-gated CI pipeline, and a tamper-evident hash-chained audit trail. Provider-agnostic (mock/Anthropic/OpenAI), FastAPI, Bicep IaC for Azure Container Apps.

SelfGPT: secure multi‑provider LLM chat with PDF RAG, web search, and LangSmith tracing.

Mystery AI Coach is a Discord AI assistant I built to help improve social confidence and communication skills. It learns from past interactions, analyzes conversations, tracks progress, awards XP for real-world actions, and provides personalized coaching and feedback. 🚀

Open-source, local-first desktop AI client and MCP host. Bring your own key for Claude, GPT, Gemini, Ollama, or any OpenAI-compatible endpoint — conversations stay on your machine.

Universal payment bridge for AI agents. Protocol-agnostic middleware supporting Coinbase x402, Google AP2, and OpenAI/Stripe ACP.

Valid tool calls from any local model — a drop-in OpenAI-compatible proxy for Ollama that guarantees well-formed tool calls and restores tool_choice.

Code Turtle "🐢" is the open-source AI code review platform that helps teams ship better code with intelligent pull request reviews, inline feedback, and noise-free re-reviews 🚀

Local multi-protocol LLM API tester for OpenAI-compatible, Anthropic & Gemini endpoints · 本地多协议 API 测活工具

Go library that cleans malformed/polluted LLM output (think blocks, tool-call wrappers, fences, preamble/postamble, truncated JSON) — stdlib only, thread-safe

检测中转站的 DeepSeek 是满血还是野鸡：用 DeepSeek Harness minimal 预设的首轮请求面做思维链行为指纹。支持 OpenAI / Anthropic / Responses 三种协议，零依赖，带网页界面。Detect downgraded or rebadged DeepSeek models behind API relays.

Open-source financial research agent whose false-positive rate is measured, not assumed. Permutation tests, multiple-testing correction, and a published calibration study that found (and fixed) a 45% false-positive rate.

终端里的编码 agent：接任意 OpenAI 兼容模型，任务计划/子代理/长期记忆/断网沙盒

Curated 2026 conference papers on Claude Code and Codex CLI—benchmarks, methods, failures, exact models, and auditable evidence.

A personal assistant that runs on your own machine, does real work on it, and writes itself new abilities when the ones it has are not enough.

An agentic coding workspace that will not claim success it cannot prove — every change goes through the project's own verification before it is reported as done

Evidence-first deep research skills for AI agents, with source tracking, citations, contradiction checks, and uncertainty-aware synthesis.

ADA Creative Co. Agency OS delivers 20+ AI roles over MCP (Model Context Protocol) and a retro Pixel Office Web Visualizer.

Run Qwen3.8 27B-parameter reasoning model with 128k context on a single RTX 3090 or 4090, locally, with an OpenAI-compatible API.

An LLM web-chat scraping tool that mimics API functionality, supporting a wide range of Large Language Models. 一个模仿API功能的LLM网页聊天文本抓取工具，支持多种大语言模型。

Knowledge-base agent demonstrating local embeddings, retrieval, tool use, and OpenAI-compatible loops.

Self-healing prompt engineering. Autonomously test, converge, and translate prompts across 447 AI models.

Autonomous terminal coding agent - Rust, ratatui TUI, OpenAI-compatible providers

Go-first framework for dependable AI agents: typed dependencies and outputs, explicit tools, composable models, observable runs

Web API sample with Microsoft Agent Framework, Azure OpenAI, and SQL Server that adds persistent per-user memory to AI agents through authenticated chat endpoints and automatic memory extraction.

OpenAIot 工业物联网中台，完整承载设备从生产、运维到客户使用的全生命周期管控。作为新一代轻量化开源物联接入底座，平台实现全品类设备统一接入；搭载零代码配置规则引擎，业务逻辑无需编码即可灵活编排，同时提供低代码开发能力，快速搭建定制化物联网数据可视化大屏。

Under construction.

Self-hosted AI operations board: run Grok, Claude and OpenAI agents as real jobs on your own VPS. Apache-2.0.

🧠 Persistent memory for AI agents. Scan logs, extract learnings via LLM, update long-term memory. Review, reflect, remember.

Agent skills for Cursor & Claude Code: goal-mode keeps agents working until tests/lint/CI are green (open /goal). npx skills add shenwell/ai-agent-skills

Storage-backed file tools and Markdown knowledge-graph context for Microsoft Agent Framework

Self-evolving personal AI agent. Telegram-first. Multi-provider BYO key. Persistent memory, daily journal, skill graph, dream cycle. Free forever. Apache 2.0.

A complete, hands-on tutorial for building AI agent skills: SKILL.md structure, deterministic scripts, templates, memory, testing, and marketplace publishing.

maok A.I. connects your local Riot Client to any AI model — including LM Studio, Ollama, vLLM, or the official OpenAI API. Whenever your League friends send you a chat message, maok A.I. intercepts it in real-time and responds directly as Maokai — wise, ancient, and protective of the forest!

Maximalist coding harness

Codex CLI is the best agent harness, but it’s locked to OpenAI. Pyxis won’t be.

Natural-language tool calling for LLM agents — Python implementation of the NLT paper (arXiv:2510.14453). Boosts tool-calling accuracy on weak/free-tier models by up to 14%.

ACE — Agentic Context Engineering: evolving, self-improving context playbooks for LLM agents. Faithful ICLR 2026 implementation with OpenAI Agents SDK support.

飞虹 Code（fhcode） — 终端 AI 编程智能体，对标 Muse Code。多模型路由、企业级 RBAC/审计/多租户、全自动 SWE Agent、离线私有化。npm 一行安装，MIT 开源。

Agent skill + CLI/library that routes agentic AI tasks to the right reasoning effort (minimal→xhigh) while keeping the model fixed — for Codex CLI and the OpenAI Responses API, with auditable escalation.

Self-hosted AI assistant with web and email chat, sandboxed approval-gated file tools, file-based memory and cron scheduling. TypeScript, zero framework.

https://youtu.be/u8KJYQgUOs8 AI-powered paralegal assistant that extracts, summarizes, and analyzes legal documents. Upload PDF/DOCX contracts to get instant summaries, extract key clauses, and perform intelligent document analysis using OpenAI.

Local AI assistant (Jarvis-like) for desktop automation, file management, and coding tasks using FastAPI and LLMs.

Visual AI workflow orchestration desktop app for MCP connect, execute. 100% offline. Supports stdio / SSE / Streamable HTTP. Built with Tauri + Rust + React.

A GOST Rewriter HTTP plugin that converts bidirectionally between OpenAI Chat Completions and Anthropic Messages API formats

A working AI agent built in C from scratch — no frameworks, no SDK. ReAct loop, tool system, filesystem sandboxing, 200+ tests.

🪶 Quill — Open-source AI super-agent framework with sandboxed execution, sub-agent orchestration, MCP, skills marketplace, session forking, FTS5 search, and a Tauri desktop app. Alternative to OpenAI Codex, Cursor, and Claude Code.

Full-stack multi-agent studio for custom character agents, streamed group conversations, tool permissions, and persistent run history.