Code agents

A quantum prompt playground & macOS-native Rust coding agent: morphing fluid entropy into rigid execution checkpoints. Entropy cannot lie.

Claude Code 技能：客户端软件全维度防破解安全审计（7大通用维度 × 9个平台专项）

🐎 hermes-agent 中文教程：38 章 14 万字，让「马」替你 24 小时写代码，配套 U-Hermes 马盘。Chinese ebook tutorial for Hermes Agent (NousResearch) with 38 chapters.

JetBrains/PyCharm plugin to review an agent's changes in your IDE and relay batched, line-anchored comments back into its running session — instead of retyping feedback into the terminal.

Official DSers MCP Server documentation and metadata for AI-powered dropshipping automation.

The OKF-go provides a high-performance suite of utilities to validate bundles, harvest metadata from databases and APIs, assemble context for Large Language Models (LLMs), and run an MCP Server.

Run opencode (open-source AI coding agent) natively on Android via Termux. aarch64/Bionic build with a patched Bun runtime — one-command install.

LLM-powered web scraping platform. Build reusable scraping services as step-graph state machines via an AI-guided wizard, expose them through a local HTTP API. Chrome Extension (MV3) + Node.js native messaging host. Data never leaves your machine.

Portable Rust Agent Client Protocol (ACP) server with model routing, tools, permissions, sandboxing, sessions, and MCP integration.

The open source agent.

Complete guide to run Claude Code CLI & VS Code with AgentRouter using a WAF bypass proxy

A deterministic control harness for AI agents. Know when to accept, retry, replan, or roll back.

Verification-first Claude Code harness by a mechanical design engineer — 13 validation gates, trust grades, heterogeneous-model cross-verification, supply-chain INTAKE, claim ledger

Autonomous pipeline that finds energy and CO₂ hotspots in any codebase and ships efficiency fixes proven by embedded statistical A/B testing, correctness gates, and integration tests reverse-engineered from the real system.

Code for 'Will Systems of LLM Agents Lead to Cooperation?' (AAMAS 2025)

Local-first checkpoints and crystals for long-running AI coding agent sessions.

A curated literature map of AI agents for robot planning, programming, orchestration, skill discovery, and self-improvement.

git diff for your agent's context window. See exactly what your LLM saw — turn by turn, block by block.

RAG for Vietnamese traffic law, FAISS retrieval + live web search behind a 3-way LLM router, with OCR ingestion for scanned legal documents

[STUDY] Application that turns into a Chat-Bot by entering the OpenAI API Key and Prompt.

Skills Gallery — 1,672+ AI agent skills. One gallery. 60+ tools compatible. Install via npm, yarn, pnpm, bun, curl, pip, npx. By uthuman Inc.

Claude Code plugin that routes heavy coding tasks from Claude to OpenAI Codex (GPT-5.6) as a supervised background worker — Claude briefs, delegates, reviews, verifies.

Farming Code is a self-hosted browser workspace for running and supervising Codex, Claude Code, OpenCode, and other AI coding agents.

Memory for AI agents that learns what actually helped: lessons injected at the moment of action, credited by outcome, challenged by counter-evidence. Local-first, multi-agent, test-gated.

Multi-agent AI orchestration framework for Python

Stop replaying the transcript: capsule session state + rolling memory for LLM CLIs (Claude Code, Codex). Measured: −90.3% vs no-cache / −30% vs cached replay at turn 10, reproducible bench. MIT, provider-agnostic.

Wolbarg — modular, provider-agnostic semantic memory SDK for AI agents

Claude agent skill that plans multi-step work as a dependency graph. Subagent fan-out, layered fan-in, model tiering, verification retry loops, human approval gates. No framework, just instructions.

The zero-trust framework for autonomous AI agents.

A fast, model-agnostic coding agent for your terminal — built in Rust with a rich TUI, MCP, lossless context, safe tools, and built-in verification.

Several Claude Code / Codex conversations in one repo? Each terminal pane resumes its own session after a reboot, not a picker. Agent CLIs scope sessions to the directory; agent-resume adds per-pane binding on top of tmux-resurrect.

Stateless CLI giving AI coding agents persistent identity, a typed lifecycle, and crash-recovery. Self-hostable, AGPL-3.0.

Self-hosted AI agent that researches US market news on a schedule and publishes structured briefs to a terminal-style dashboard. Bring your own LLM.

Failure Compiler for AI agents: turn failed outputs into replayable regression cases.

Extends Unix tree so every file shows its one-line concern and boundary, giving a separation-of-concerns overview of the whole workspace at a glance. A fast, enforceable map for AI agents.

Scaffold and manage plugin marketplaces for Claude Code · GitHub Copilot · OpenAI Codex · Cursor, distributed through any Git host

🕹️ Gamified Vibe Coding, with full controller support, mini-games, themes, and more!

Reusable agent skills for battle-tested engineering judgment — least code, fix the problem, don't break what works. Works with Claude Code, Cursor, and Codex.

AI-powered research reasoning agent for research gap discovery, novelty assessment, and implementation guidance.

A super-fast, ram-efficient, and lightweight agentic code editor that's ready to ship your code to the next level!

Reliability & security proxy for the Model Context Protocol (MCP). Self-healing connections, runtime tool-poisoning/shadowing defense, and NIST AI RMF + OWASP LLM Top-10 audit trails for any MCP server. Works with Claude, Cursor, Cline, Windsurf.

Your personal Claude assistant, running on OnCell — a NanoClaw fork where every agent lives in a durable cloud cell instead of a container on your laptop. Isolated, crash-proof, remembers everything, costs ~$0 while idle, wakes on your next message.

Self-hosted AI agent you talk to on Telegram. Docker, full tool access, 850+ skills, voice in and out, layered security.

Self-hosted EVE Online AI assistant for Telegram, Discord, and CLI — EVE SSO/ESI, local SDE, and OpenAI Responses API.

🎅 A Claude Code skill that reads its memory of you and builds you a surprise gift while you sleep. One command: /santa

Open-source AI engineering team. Prompt in. PR out. Specialized agents plan, build, review, test, and ship code — with human approval at every gate.

Cut your Claude API bill: local proxy that auto-routes every request to right model. Zero code changes.

High-performance Rust proxy that translates Anthropic Messages API to OpenAI-compatible format, routing to NVIDIA NIM, Ollama & OpenRouter. Built with Tokio + Axum.

Personal collection of Claude Code plugins and skills: development workflows, cross-model review, issue-to-PR automation

Local operations hub for Claude Code — a grid of real claude CLI sessions with live working/waiting status, usage & cost tracking, multi-account profiles, and revivable sessions. No API keys, no cloud, $0.

AI-driven incident response toolkit. Reconstructs the full attack path of a cloud or Windows breach, maps every step to MITRE ATT&CK, and stages human-approved containment. AWS · Azure · GCP · Windows. Apache-2.0.

養成型 AI 伴侶框架｜Raise an AI companion who remembers, reflects, and grows with you.

Single-binary LLM gateway (gw) in Rust — OpenAI/Anthropic-compatible APIs, multi-provider routing, per-key auth/quotas/rate-limits, failover, streaming, and a billing ledger

Your AI's personal travel agent — 66 tools, zero API keys. Flights (incl. fares airlines hide), hotels, visas, weather, points & miles, via MCP.

Searchable, portable store for your Claude Code sessions — full-text search, AI titles, tags, fuzzy find, and resume any session in any folder.

A local Claude Code dashboard for personal analytics: track token usage, API-equivalent cost, sessions, tools, skills, projects, and activity patterns.

Desktop app for managing multiple Claude Code sessions - graph view, split groups, persistent memory, todos, inter-session messaging

Your agent can write the change — it can't tell you what else depends on the code it's touching. This builds a model of your product from your own code, then scores every task: what it changes, what that reaches, which user journeys and sensitive data are in the blast radius — arithmetic on screen. Local, zero deps, no telemetry. MCP included.

Workflow-agnostic guardrails for AI coding agents: spec sync + risk-triggered engineering discipline for Claude Code and Codex.

Give your AI agent a memory. And a seatbelt.

Deterministic AI agent orchestration — comment via Discussions & Issues; code changes by maintainers only.

Coroutine-first Kotlin client for the Qdrant vector database: suspend APIs and type-safe DSLs, REST or opt-in gRPC behind one seam, and a multiplatform core.

A Query-Local Unified Risk Modeling Framework for Corpus Poisoning Attacks

An end-to-end AI-powered mobile productivity assistant that transforms natural language into reliable actions for managing tasks, meetings, notes, and progress. Built with intelligent language understanding, deterministic execution, and workflow automation for efficient productivity management.

Turn any markdown vault or codebase into a cited, AI-queryable knowledge graph. Local-first CLI, web viewer, and MCP server.

Local-first personal Markdown knowledge base with semantic search, AI Q&A, MCP, desktop/web clients, and a self-hostable zero-data-at-rest relay.

Open-source visual AI workflow platform powered by Skills, MCP, agents and extensible components.

A safety preflight tool for Claude Code and Codex that decides whether a coding task should loop, then generates a bounded execution contract.

Collaborative travel planning SaaS — plan itineraries on a map, vote on stops, and split expenses fairly.

Local-first nested-learning AI agent runtime with Memvid v2 memory, safe tool use, and a web workbench.

Template for building protoLabs A2A agents on LangGraph — fork me

Website visual cloning skill for Claude Code — 6-phase process: recon, foundation, spec per section, parallel build, assembly, screenshot QA.

Plan-and-Execute AI Coding Agent with Verification & Safe Tool Orchestration | 规划-执行型 AI 编程助手

git blame tells you who. gitwhy tells you why — provenance for AI-written code (Claude Code + Codex), 100% local.

AI Workforce plugin for Claude Code — proactive sales & marketing strategy for startup founders. 24 domain knowledge skills, 10 commands, 4 AI agents. Integrates 15+ strategic frameworks.

Unified memory and skill infrastructure for AI agents — pull-based ingestion across Claude Code, OpenClaw, Codex, Cursor, and more. Markdown canonical + SQLite hybrid index + MCP-native API.

An agent composition framework that builds stacks and compiles specialized subagents for Claude Code

Study guide for all four Anthropic Claude certifications: Associate, Developer, and Architect Foundations and Professional. Official exam guides, blueprints with domain weights, cheat sheets, 320 practice questions, three mock exams per exam, 110 flashcards, and a printable companion.

Local-first AI assistant with real memory, agent tool use, and proactive habits — built with Go + Flutter. Runs fully offline via llama.cpp, recalls context across sessions with local RAG, edits files/runs commands as an agent, and integrates WhatsApp + calendar. No cloud, no telemetry, open source.

Persistent per-project memory for Claude Code, Kiro, Cursor & Codex - plain markdown committed with your code, captured prompt-free, auto-injected at session start, recalled by meaning (local semantic search, zero API). Cross-project persona, cited recall, searchable session history.

Let AI agents run, test, and debug inside your Okteto dev environments

Choreographic coding agents runtime

Cognitive OS v3.0 — Universal AI agent operating system specification. 16-section cognitive architecture for Claude Code, Cursor, Codex, Gemini, and Factory Droid. Multi-harness prompt engineering with verification chain, swarm coordination, and adversarial security.

AI advisory board as a Claude Code skill + MCP server: verifies every quote word-for-word against public-domain sources, or abstains. Fail-closed fidelity.

Adversarial security benchmark for agent authorization: does a compromised agent's policy-violating proposal become an unauthorized external effect? 73 trials, nine families, an independent oracle, per-mechanism ablation, confidence intervals. 0 unauthorized effects in 61 attack trials (95% CI [0.0%, 5.9%]). Reproduction is partial.

A Python library for evolving code and algorithms using Large Language Models (LLMs)

AI Engineering 实战学习仓库

AI-NewsPapaer is an intelligent, autonomous Telegram bot designed to serve as a personal news desk.

Praison is a production-ready framework for building multi-agent AI systems that enables intelligent agents to collaborate, reason, self-reflect, and use tools to automate tasks, with support for over 100 LLMs and flexible workflows for both code and no-code usage.

Lightweight observability for multi-agent AI pipelines. Self-hosted, no cloud dependency — add two lines of Python and see real-time agent events, DAG graphs, token usage, and session history on a local dashboard.

Multi-agent orchestrator — route tasks across Claude, Codex, Gemini, agy, opencode. Model selection + escalation + output aggregation.

A tiny native macOS menu bar app for monitoring Codex usage - no browser extensions, no Chrome access, no Keychain access, no telemetry.

Reliable, resumable, code-driven orchestration for sub-agents and agentic-loops

Subagents that remember — role-specialized AI personas with persistent per-project memory, plus a live lead-and-crew channel (dispatch, steer mid-task, human-in-the-loop approval) for Claude Code.

企业 AI 中台 — 让 AI 连接一切 | Enterprise AI Platform with Agent, MCP, RAG, Plugins

Inja LLM Gateway — stateless OpenAI/Anthropic API gateway with multi-provider routing and usage hooks.

Use Claude Code 100% free with 100+ NVIDIA NIM models via LiteLLM proxy. No Anthropic subscription needed. Works on Windows & macOS.

One-shot simulations: real physics & math, each written in a single attempt by Claude Fable 5 — one self-contained HTML file, no frameworks, no fakery

繁體中文寫作 skill 工具鏈 for Claude Code：humanizer-tw（去 AI 味）+ good-writing-zh（節奏打磨：PG／余光中／王鼎鈞 + 技術文件保守模式）+ fable-econ／fable-explore（Amanda Askell 寓言式概念學習）

On-demand Claude Code skill for AI/ML news, tools, papers & learning resources — search-first, cited, dual-layer (ELI5 + technical + verdict), role-aware.