Openai agents

轻灵 Qling - 本地优先的中文 AI Agent CLI 工作台 / Local-first Chinese AI Agent CLI workbench

Dependency-free static analyzer that audits LLM agent tool/function definitions (OpenAI, Anthropic, MCP) for excessive agency, code-execution & SSRF sinks, secret access, and data-exfiltration surface — mapped to the OWASP 2025 LLM Top-10. Runs in CI.

Loopback bidirectional Anthropic/OpenAI API proxy with tool-call validation and multi-provider routing.

OE Runtime - run AI agents against enterprise data sources. One YAML agent, one config file.

Klyrune — agentic terminal assistant with multi-provider support (OpenAI, Anthropic, DeepSeek, 9Router, custom endpoints), tools, sessions, memory, todos and plan/build modes.

TypeScript agent orchestration entrance (createAgent, createTeam) — durable memory/HITL/stream replay when you need them. Production-oriented architecture, young ecosystem — not a graph DSL.

Noir model — optional bounded single-shot LLM layer (anthropic / openai / openai-compatible); agent loops are impossible by design.

A small, provider-neutral TypeScript toolkit for calling LLMs from backend code.

Cross-platform OpenAI-compatible Qwen coding agent CLI

A lightweight coding agent with OpenAI-compatible and Anthropic Messages API support.

Universal context-compression proxy for AI coding agents. Sits between any agent and its model API, rewriting Anthropic/OpenAI streams with acp-kernel compression. Any agent that can set a base URL (Claude Code, Codex, Cursor, Aider) works out of the box.

OpenAI Agents SDK-backed AgentHarness implementation for @salesforce/sfdx-agent-sdk

Turn a supported ChatGPT/Codex OAuth sign-in into a private OpenAI-compatible endpoint, starting with self-hosted n8n.

OpenAI fast mode toggle for pi - Enables priority service tier on supported GPT-5 and GPT-6 models

A lightweight terminal chat agent built on the OpenAI Responses API through the official @eliware/openai client.

looprun model server: expose governed LoopRunAgents behind an OpenAI-compatible /v1/chat/completions endpoint, so any harness that speaks the OpenAI protocol can call a governed agent as if it were a model. The full governed turn (guards, tools, redrive)

Vercel AI SDK provider adapter for openai-oauth auth handles.

React voice agent UI components with audio-reactive adapters for Vapi, ElevenLabs, LiveKit, Pipecat, OpenAI Live, OpenAI Realtime, Gemini Live, and custom voice AI apps.

OpenAI and OpenAI-compatible AI provider adapters for AgentProdReady (chat, tools, streaming, embeddings).

TypeScript runner for Sovara — wrap a script in a single function to record every LLM call, tool invocation, and log line into a structured run graph.

Local Codex auth-file auth handle for openai-oauth.

Agent SDK for NEURA Compute — OpenAI-compatible inference with automatic HTTP 402 -> burn $NEURA -> retry handling, and wallet-only (SIWE) onboarding for autonomous agents.

Offline BPE tokenizer for OpenAI, Anthropic, and Gemini — zero dependencies

OpenAI model stream adapter for AIGUI.

Production-ready AI Assistant Server with advanced RAG (chunk-based semantic search + adjacent retrieval), conversation management, real-time communication, and human agent handoff

**Multi-Agent AI Assistant built on OpenClaw**

Just A Rather Very Intelligent System - A developer-first, type-safe, transparent multi-agent SDK

Umbrella package for the Agent Framework (TypeScript): one install for the core, the OpenAI / Anthropic / Foundry providers, MCP, A2A and hosting, each re-exported under a subpath.

A from-scratch TypeScript agent SDK - agent loop, tools, guardrails, sessions, handoffs, structured output, streaming, and tracing. No agent framework dependency.

An open-source AI Agent framework in TypeScript with tools, memory, and guardrails support.

XRLF Protocol — Universal OpenAI-compatible proxy that gives any LLM foveated ring memory, active recall, and cognitive steering. Point LM Studio, Ollama, or any OpenAI SDK at http://127.0.0.1:8300/v1 and it just works.

Type-safe LLM-powered functions. Write intelligent functions with decorators, not agents.

Approve or deny your AI coding agent's risky commands from your phone, with an audit trail. Works with Claude Code and OpenAI Codex.

AI Dynamo LLM serving client — OpenAI-compatible inference client with multimodal support

OpenAI provider for BindAI

Caesar: autonomous AI research agent with graph-based deep web exploration and adversarial answer synthesis.

OpenAI-compatible API server for the Claude Code CLI — streaming, tools, vision, and live session reuse.

CLI AI coding agent: ReAct tools, hybrid RAG, multi-provider LLMs

Deterministic Adversarial Simulation (DAS-LLM) Framework for LLM Security Invariants

A native Python port of kiln — build and orchestrate AI agents: unified providers (OpenAI/Groq/Ollama), a tool-calling loop, self-correcting structured output, MCP support, and a DAG orchestrator.

OpenAI agent handler for LaunchDarkly AI SDK

OpenAI messages handler for LaunchDarkly AI SDK

Native OpenAI-compatible agent harness for LiteyukiBot v7.

llama-index llms TrustedRouter integration

Resilient Multi-Provider LLM Fallback Router (DeepSeek, OpenAI, Ollama, Anthropic).

OpenAI Agents SDK invocation for ml-pipes

OpenAI Agents SDK invocation for ml-pipes

OpenAI Responses API invocation for ml-pipes

Mutiny Adapter #1 — OpenAI Agents SDK TargetAdapter

Use OpenAI, Gemini, and any OpenAI-compatible model inside the Claude Code CLI.

RunAPI OpenAI Transcription SDK for speech-to-text integration

A modular, extensible AI Agent framework with multi-provider support and plugin architecture

Standalone Real Memory AI Agent CLI for project chat, coding workflows, provider configuration, and safe Git operations.

Minimal tool-loop harness for LLM agents. Keep your provider SDK, keep your messages, add a correct tool loop.

OpenAI provider plugin for Shiro.

Show OpenAI Codex, GitHub Copilot, Z.ai GLM Coding Plan (global and China), and DeepSeek balance usage in the pi coding agent.

AirRoute — free AI gateway. One endpoint, many free providers, auto-fallback, adaptive auto routing, token compression, and a full admin console. Works with Claude Code, OpenCode, Cline, Copilot & any OpenAI/Anthropic-compatible agent.

Stopgap opencode plugin that smooths the OpenAI/Codex GPT-5.6 rough edges (Luna 404s, missing session titles, reasoning-summary markers) until native support lands

SamAI SDK is a provider-agnostic TypeScript AI SDK for building AI agents, chatbots, tool calling, RAG, MCP, streaming, voice agents, and multi-provider LLM applications using OpenAI, Claude, Gemini, Groq, Mistral, Ollama, Azure OpenAI, and AWS Bedrock.

Headless React primitives for AI chat — state management, streaming adapters for OpenAI, Vercel AI SDK, and AG-UI, message format converters, and thread management for OpenUI generative UI apps

Abstractflow — zero-config JavaScript AI Agent SDK with tools, guardrails, multi-provider models, and multi-agent handoffs. Install, drop your key in .env, run.

Ayu Agent Sdk a from-scratch, provider-agnostic TypeScript AI Agent SDK with tools, handoffs, guardrails, memory, structured output, streaming, and tracing.

FastMCP server that delegates work from Claude Code to OpenAI Codex CLI and Google Gemini CLI as headless subprocesses.

A unified LLM wrapper for online APIs and local models

通过openai兼容的api代理运行claude code, 不修改claude code的配置

Python toolkit for AI-assisted applications - an OpenAI-compatible LLM client, text processing, web scraping, and interactive demos

OpenAI Agents SDK recording adapter for Kitaru.

Provider-neutral synchronous chat adapters for OpenAI, Claude, DeepSeek, OpenRouter, and LiteLLM.

Share an LLM API quota across devices on a local network (Wi-Fi hotspot, USB tether, or LAN)

CLI tool for OpenAI-compatible APIs via AceDataCloud

A local coding and terminal AI agent powered by an OpenAI-compatible Llama server.

The OpenAI Agents SDK (Python) adapter for Welt's wire contract

AI Agent Kit for 1inch - Connect any LLM to 1inch DeFi protocols

The official ToolPlex client for AI agent tool discovery and execution

Multi-provider language generation toolkit using Vercel AI SDK - generate responses with 10+ LLM providers including OpenAI, Anthropic, Google, and more.

LLM-agnostic AI platform for SAP CAP. 11 providers behind one API (Anthropic, Google Gemini, AWS Bedrock, Azure OpenAI, Ollama, Groq, Fireworks, DeepSeek, Mistral, OpenAI-compatible, SAP Generative AI Hub) with 30+ middleware primitives spanning cost/resi

OpenAI-compatible HTTP server over ACP agents. Use Claude Code, Codex, opencode or any ACP CLI from anything that speaks the OpenAI API but does not speak ACP yet.

A lightweight OpenAI agent harness with tools, builder API, and a ReAct-style JSON pipeline.

Local x402-paying proxy + MCP server for openzoo.fun — point any OpenAI-compatible harness (Cursor, Claude Code, aider, SDKs) at localhost and it pays per call from a local burner wallet. Solana and Base rails live; Robinhood experimental. `openzoo build|

Zero-config agentegrity adapter for the OpenAI Agents JS SDK (@openai/agents). Mirrors the Python agentegrity.openai_agents module.

OpenAI provider for chai-ai.

Minimal Pi extension for OpenAI Fast mode.

This is the OpenAI provider for the Rabbit Agent SDK.

Improve OpenAI in pi with fast mode, usage stats, realtime voice, image generation, and footer polish.

Vite plugin for packaging OpenAI Sites deployment metadata.

AgJSON normalizer for the OpenAI Agents SDK (RunItemStreamEvent / Responses streaming → AgEvent[]) — fixture-tested.

OpenAI-compatible chat completions provider adapter (OpenAI and any compatible endpoint) with streaming and tool calls.

OpenAI model provider adapter for @purista/harness.

Runtime control, proof, and fleet intelligence for MCP-compatible AI agents.

OpenAI SDK adapter for llm-ports. Implements LLMPort and EmbeddingsPort. baseURL support covers 10+ OpenAI-compatible providers (Azure, Groq, Together, Fireworks, DeepInfra, Perplexity, Cerebras, LiteLLM proxy).

ANTRI Code - The Autonomous Agentic CLI & Desktop Assistant

Multi-agent, multi-provider harness for building software autonomously with any LLM. Researcher, Planner, Curator, Generator, Evaluator, Documenter pipeline with a fail-closed security-audit gate, a fleet orchestrator for bulk multi-agent runs, and a cros

Standalone, multi-provider AI Agent framework in plain JavaScript

Model-provider-agnostic agentic CLI — npm installer for the Python distribution

Bridge OpenAI tools to Claude Code SDK, Codex CLI, and OpenRouter

Provider-free Runner regression probes for the OpenAI Agents SDK

Agnos Proxy - the gateway-agnostic control plane for LLM routing, cost-tracking and observability. OpenAI-compatible governance proxy over a swappable BackendEngine.

OpenAI Agents SDK sessions backed by Citadel: encrypted at rest, with deletes that destroy the key

Multi-agent collaboration plugin: broadcast, sequential, debate, supervisor, consensus and relay strategies over DeepSeek/OpenAI-compatible LLMs

Budget checks for OpenAI API costs, locally and in GitHub Actions.