Code agents

Open-source AI coding agent for VS Code — plans, edits, and runs tasks with your approval. Bring your own model: OpenRouter, or a local one via Ollama, vLLM, or LM Studio.

AI-driven global industrial policy tracking skill for Claude Code.

Aria. A multi-agent digital marketing assistant. Give it a business objective and a brand tone; a team of agents writes the copy, plans the campaign, scores itself against a rubric, and reworks until it passes. Model-agnostic via a gateway.

LLM Tool for effective test evaluation of ML projects with curated Checklists and LLM prompts

Two Claude Code Skills for Codex collaboration: execution delegation discipline + plan co-creation discipline, avoiding context pollution and false second opinions in dual-model workflows

Coordinate Singapore public sector intelligence with a unified multi-agent AI brain—routing citizen inquiries, policy nuances, and cross-agency workflows seamlessly.

The Everything Prediction Engine — five AI analyst personas debate, 10,000 Monte Carlo worldlines unfold, and superforecasting calibration converges any question into one actionable probability.

Shared memory MCP server for multi-agent projects. Agents (Claude, Cursor, Codex, AntiGravity) write to a local `.agent-mem/` folder so the next agent picks up where the last one died.

Local-first, Codex-native agent foundry and knowledge OS for creating reviewed, testable specialist AI agents.

MT5 Auto-Backtest & Optimization Pipeline — Headless CLI for strategy evaluation with risk-adjusted scoring

Keep a Claude Code task going across usage limits — waits for the reset and resumes the same session. Holds your Mac awake while it waits.

A small, self-hosted MCP server that gives Claude read/write access to your Google Sheets, optimised for analytics on tabs of 50k+ rows

llama.cpp gguf gemma local-ai agentic-ai coding-assistant

Graph engineering protocol for Claude Code and Codex teams: spec freeze, durable roles, isolated workers, independent review, and evidence-backed delivery.

Orphus — an agent harness whose agents deliberate in rooms outside their context windows. Fork of Atomic (Pi lineage) plus bounded multi-agent discussion, budgeted digests, and a declarative role manifest.

Adversarial multi-agent code review — a committee of specialist reviewers that catches what single-pass AI review misses.

A calm home for coding agents, terminals and projects — local-first, persistent, and reachable from any device you trust.

Cross-harness orchestration for wakeable coding agents using Agent Intercom

Six MCP servers that automate the full academic research pipeline — from refining a vague research question to generating a publication-ready report. Each server handles a distinct stage of the workflow: question development, data processing, code generation, script execut

Claude Code plugin marketplace. audit: a manifest-driven audit/fix pipeline with enforced plan-first and TDD gates — deterministic hooks, not prompt suggestions — token spend attributed to phases and tasks, and a shareable HTML report. Zero dependencies.

Persistent memory scaffolding for AI agents - Trail, Destination, Improve, Retrospect, Intent, Probe. Self-built across 221 iterations; declared complete only when GPT, Claude, and Gemini independently found nothing left to change.

gguf safetensors phi3 ai-model-builder fine-tuned llama.cpp

mlx safetensors qwen3_5 text-generation mlx-mtp qwen

DeepSeek Harness Desktop 是 @KevPH2026 维护的非官方社区 macOS 桌面封装:基于开源 DeepSeek Harness,内置一键安装的精选社区皮肤中心、Telegram 手机远程控制、可配置的图片/视频多模态生成、用量面板和按场景分类的 dsh 插件市场。同步社区雷达,未签名 CI 发版。

Universal zero-token AI gateway: LLM plan-cache + real device control (Android via rish/Shizuku or ADB, desktop shell)

Context Excavator solves a real problem: when you paste large codebases into ChatGPT, the AI gets lost in noise. You're feeding it 500 lines of code when only 20 lines actually matter. The LLM can't tell the difference, so it gives irrelevant answers. This project extracts only the code that matters, automatically..

Rules your coding agent physically cannot ignore. Claude Code hooks that DENY the tool call and BLOCK the reply - stopping reactive patches, guessed citations, and premature 'done' claims. Also works as an LLM-agnostic rule pack.

Run Claude Code on your Windows PC from Discord -- multi-session channels, local voice in/out (drive mode), git worktrees, and confirm-gated dangerous actions.

chat with an llm about an arxiv paper by using the latex source

Multimodal Telegram AI bot: image understanding, voice STT (Whisper) and TTS voice replies, smart group-chat moderation, persistent Gist memory. Gemini or any OpenAI-compatible LLM (Groq, LiteLLM, OneAPI). Single Flask file, 3 dependencies.

AIML Chat bot written in node.js for Discord.

NKUST LINE Bot 是基於 LINE Messaging API 進行開發的 LINE Bot，透過該服務整合了學校的校務系統資訊，可以在 LINE 上直接查詢自己的課表、學期成績、出缺席紀錄與校園資訊等。

Starter kit for educators to test, adopt and iterate on genAI tools in the classroom

A concept face recognition featured demo of virtual police station where people even in remote places with internet access can get help while using it as home security for potential criminals or missing people are detected by your existing home surveillance system.

Bobby is the coolest mf

A powerful voice-activated chatbot using Google Generative AI and Google Search API. It leverages large language models (LLMs) to listen, respond, and search the web for you.

🥇 2nd Place | Solution for Hackathon Kiln 2025 | AI-powered assistant that both educates on blockchain concepts and can help advanced users quickly retrieve precise, real-time data using Kiln's API

OrbitBot é um bot de WhatsApp com IA projetado como um sistema completo de automação e gerenciamento de conversas. Combina respostas inteligentes com contexto contínuo, banco de dados otimizado em SQLite, sistema de filas, cache multi-nível e monitoramento em tempo real.

Moroccan Meal Recipes Chatbot Using Rag technique, with Ollama, Openrouter API, Faiss DB and Streamlit

Welcome to RootTron - The AI-powered cybersecurity assistant. Expert guidance in ethical hacking, penetration testing, and security automation. 🛡️💻

Upload a CAD PDF to extract text and automatically generate a concise engineering summary using a local LLM.

A production-ready MSA blog template for beginners. Just fill in your .env and launch your own observable microservices ecosystem.

This is a clone of the OpenAI chatGPT website that instead uses the GPT API

LLM leveraging OpenAI API to be used in Glific via webhooks

A starter kit for building realtime speech-to-speech applications using the OpenAI Agents SDK and the GPT Realtime API on Azure AI Foundry.

Advanced AI-OCR with FastAPI and OpenAI Integration

🎭 Talk to Claude in 100% emoji. A stylized communication mode for Claude Code.

Local task queue & scheduler that squeezes Claude's 5-hour usage windows dry — and now any Anthropic-compatible subscription (Kimi Code, GLM Coding Plan, MiniMax, MiMo, OpenCode Go, Ollama Cloud) via engine profiles, each with its own cooldown and ledger. Single Go binary, zero-quota orchestration, read-only web board.

PM-focused 12-skill curated subset of claude-skills. For product managers using Claude Code.

Benchmark LLM jailbreak resilience across providers with standardized tests, adversarial mode, rich analytics, and a clean Web UI.

Create, validate, and maintain project-specific Claude Code harnesses through a structured interview.

CLACOROO, the Claude Code Control Room: free desktop app that shows what eats your context window and quota, browses sessions, and manages plugins, marketplaces, skills, agents, MCP servers, hooks and permissions through the official claude CLI. macOS, Windows, Linux. AGPL.

22 Claude skills for game developers. Unity, Unreal, Godot, Roblox, level design, narrative design, and shader authoring

Two-line status line for Claude Code: which checkout you're in, context remaining, and live session cost. Pure bash + jq — one file, no dependencies.

Production-grade multilingual RAG for scientific research across 12 languages (English + 11 Indic). Uses BGE-M3, ChromaDB, hybrid BM25+dense retrieval, cross-encoder reranking, and Gemini 3.6 Flash. Includes fast RAG, LangGraph agents, 6 research tools, NLI verification, multimodal indexing, and grounded citations (<2% hallucinations).

Multi-Model AI Chat Platform. Experience the future of AI conversation with multiple models, web search integration, and document analysis.

Git-native structural memory for AI coding agents. Index once — stays fresh on every pull, merge, and branch switch. MCP server for Claude Code, Cursor & any agent.

marketo-cli: Marketo CLI for the Marketo REST API. The installed command is mrkto.

Multi-DB RAG translation agent — FAISS translation memory, glossary, and ChromaDB style retrieval, orchestrated with a LangGraph router/critic loop. ENGLISH - SPANISH - FRENCH

An open-source engine for deciding when messy records of people, places, organisations, and products refer to the same real-world entity, with evidence, not just similarity scores or identifiers.

HelpDesk Enterprise open source: agente RAG de soporte IT con memoria autocapacitada, Gemini gratis, FastAPI + Streamlit + conector Copilot Studio

Terraform-style plan/apply for agent systems: versioned YAML for agents, tools, workflows, and policies; local-first SQLite state; MCP & HTTP tools; structured traces.

Turn rough notes into a polished, schema-validated slide deck in any AI agent — export editable PowerPoint/Keynote/Google Slides. npx @presentation-md/install · npx skills add isatimur/presentation-md

Remote-control your own coding agents. Send a task from your phone; your home PC runs GitHub Copilot CLI or Claude Code on your local repo, runs your tests, and reports back. Telegram + installable web app, zero recurring cost.

Local AI assistant proxy — connect ChatGPT/Claude/Gemini/Qwen to your local filesystem & browser via a Chrome extension + sandboxed Go server.

Supervision-first AI coding agent in Rust — visible tool calls, permission gates, reversible snapshots, and token-aware context in one 11 MB native binary. TUI + REPL, Anthropic & OpenAI-compatible providers, MCP, prompt caching.

HookRadar Creative Intelligence skill/plugin for Codex, Claude, MCP, and SKILL.md-compatible agents

Chat-driven AI agent that builds full-stack apps through an autonomous SDLC pipeline. 18 specialist agents, self-learning loop, any LLM provider.

Rokid AR glasses voice → your own AI brain (default Claude); glasses-native app + self-hosted backend, pluggable brain/STT/ingress. ｜ Rokid 眼镜语音 → 你自己的 AI 大脑（默认 Claude）；眼镜原生 app + 自建后端，大脑/STT/公网入口可插拔。

Multilingual voice and text conversational AI agent that automates sales calls, contextually answers questions using RAG, and dynamically segregates leads into HOT/WARM/COLD pipelines in 10 Indian languages.

🚀 基于 Agentic RAG 与 FAISS 向量检索的沉浸式英语记忆系统 | 包含多模态 OCR 结构化、词根知识图谱星状图、语义雷达与 SRS 动态复习闭环。

Cuatro skills de Claude Code para construir con agentes sin que el resultado salga correcto y aun asi inutil: SPEC antes de construir, reparto en bloques con dueno exclusivo de archivos, verificacion adversarial y segunda opinion con Codex.

Equaxis: a governed, observable and reliable agent runtime powered by Pi.

Point Claude Code at one goal and walk away. A Node control plane the agent can't bypass - a separate verifier decides when it's really done, guardrails veto every risky action, and hard budgets make "loop forever" impossible. It finishes for real or stops. main moves only when you merge.

Supercharged Agentic Analysis Swarm — Multi-agent Claude Code plugin for end-to-end data analysis automation

Static system dashboard for a Hermes Agent installation: host status, security, model routing, honest cost, memory, cron economics and a capability map — built by cron from the agent's own telemetry. Stdlib-only Python, EN/RU.

SWARM is an open-source Codex plugin for directing a coordinated team of AI agents from one control task.

Open-source workspace where AI agents talk, drive browsers, and recruit help — terminals, notes, and live browsers wired together on one canvas. Cables are the permissions; every agent speaks one CLI.

Private, self-hosted coordination control plane for coding agents

AiFP 记忆感知系统｜MCP 服务，一套记忆全 AI 共享。面向中文的 Agent 感知记忆，支持叙事链、语义纠错、感知链图扩散。兼容 DeepSeek‑Harness、Claude Code、Cursor、Codex等全部 MCP 客户端，数据完全本地存储。

A RAG agent that answers from your docs with citations, or abstains when the evidence isn't there.

Persona taste skills and harness

A lightweight, self-hosted AI characters chatbots app based on NodeJs and SQLite. Supports SSE streaming and automated history summarization, image genereation

Spring Boot + Vue 3 教务管理系统 / academic management system for graduation projects, with Redis 抢课, course selection, grades, exams, timetable, AI assistant, natural language SQL, and load testing.

transformers safetensors iquestcoder text-generation code industrial-code

A lean docs skeleton for Laravel + Claude Code. CLAUDE.md is RAM, docs/ is disk.

Self-hosted, private AI agent on a Raspberry Pi (4 or 5). Telegram chat gateway, local Chromium browser for web search (no third-party APIs), and per-user memory backed up to your own private GitHub repo. Outbound-only, no inbound ports, survives reboots. One interactive installer, local or over SSH.

22 Claude skills for game developers. Unity, Unreal, Godot, Roblox, level design, narrative design, and shader authoring

Static code analyzer for any repository — classify codebase, extract HTTP routes, detect tech stack, map dependency graph. Multi-language (Python, JS, Java, Go, Ruby, PHP). Zero dependencies. Optional LLM enrichment

Merge Claude Code & Cowork local sessions across subscription and API-routing accounts on macOS. Automatic backups, safe rollback.

A standalone, all-on-by-default token-optimization layer for LLM requests: minifies tools/system/messages (code-fence aware), dedups tool results, distills old turns, prunes to a budget, type-compresses output. Async proxy for Anthropic, OpenAI & Ollama + MCP server (stdio) + CLI. MIT; ships orjson/xxhash/tiktoken.

MCP server for YooKassa — payments, refunds, receipts, webhooks (Russia). Part of WWmcp — 46 MCP servers for emerging markets.

Test suites + deterministic Python for your agent skills' decision logic. Adversarial persona reviewers write the tests; the generated code must keep passing them — zero LLM calls at inference. Try: uvx temper-skills audit <skill.md>

A team-wide prompting coach. Scores every prompt sent through Claude Code, Claude.ai, VS Code or Copilot Chat on 5 dimensions in real time, coaches the weak ones, and feeds the strong ones into a team wiki that gets injected back into the next person's context.

Pharmacist-facing drug-interaction decision support — deterministic detection over 152K records, LLM explains only. FastAPI · Postgres · React · RAG

Chatbot open-source de orientação, saúde e apoio emocional para a comunidade LGBTQIA+. 🏳️‍🌈

safetensors qwen3 TeichAI unsloth text-generation conversational

safetensors qwen3 chatgpt codex v2 unsloth

⚡ Advanced AI iteration framework featuring an 8-stage pipeline, automated fine-tuning, voice interaction, and security scanning.