Code agents

Open a browser tab. You're now driving your Mac mini. Live remote desktop + real PTY shell (Claude Code, vim, top) + on-device voice. Self-hosted. Apple Silicon native.

gguf qwen3_5_moe_text qwen3 moe a3b mtp

从《矛盾论》《实践论》完整方法论蒸馏的中文 AI 产品经理（AI Product Manager）/ Product Decision Agent Skill：解决需求、增长、运营、数据与项目协作；支持 Codex、Claude Code、Cursor

An open-source coding agent built with Rust.

Agent that produces narrated, research-grounded explainer videos for any topic, code-driven visuals with real narration.

Autonomous multi-agent crypto intelligence & trading ecosystem for Robinhood Chain L2 (EVM #4663) operated via Discord Command Center, Terminal TUI, REST API, & Telegram.

面向中国用户的开源 Grok Bot 替代方案，默认简体中文，支持 Claude、Codex、GPT、Grok、Gemini 与 API 中转站。

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

A plug-and-play framework for AI-assisted software development, enhancing context-aware collaboration in complex codebases. Perfect for tools like GitHub Copilot. 🐙✨

A custom fork of oh-my-openagent focused on advanced agent routing, workflow safety, and project-aware automation.

DeepSeek Harness long-term memory & self-evolving plugin: L0 capture -> L1 memory extraction -> L2 scene grouping -> L3 user persona, auto recall injection + skill synthesis, fully local.

A small terminal agent for local LLMs. One Go binary, no runtime, no server. Point it at Ollama, LM Studio, llama.cpp, vLLM, OpenRouter — anything that speaks the OpenAI /v1/chat/completions format. It reads and writes files, runs commands, and keeps the conversation where you can see it.

Deeptop, a lightweight native desktop client for DeepSeek Harness.

A DeepSeek Harness (DSH) agent preset that teaches while coding — concrete scenario-grounded explanations, Socratic guidance, and TODO(你) practice blanks, modeled on Claude Code's Learning output style. 学习模式

Three Agentic Skills for Matrix: encrypted chat with a daemon that follows rooms live, Synapse homeserver administration, and content guidance for announcements. Chat works with any homeserver; administration requires Synapse.

Same image, same prompt: 15 AI-built Blender scenes across GPT-5.6 Luna, Terra and Sol, with renders, scripts, editable scenes and audited telemetry.

Hypothesis-driven problem solving for AI agents: probe, falsify, escalate — never blind-retry. MCP server + agent skill built on Newton, Popper, and NASA's P10

The agent-native systems programming language: Meaning in. Verified machine code out. All target OS, memory managed through Ownership & fully interoperable with existing ecosystems & object-oriented AI Agents with payment capability.

milk — a CLI/TUI that routes prompts between agents: a primary agent and a configurable escalation agent, with session-aware state management and real-time streaming. Escalation happens automatically (via routing rules) or manually, preserving context across the handoff between agents.

An agent package and project manager, written in Rust.

Provenance-aware runtime security for AI agents with deterministic ALLOW / REVIEW / BLOCK enforcement.

Session continuity for Claude Code: context-usage sensor, threshold alerts, HANDOFF state files, auto-recovery after compaction. EN/RU docs.

PolyForge: a Rust CLI toolkit that turns any GitHub repo into an agent-verifiable knowledge base via MCP (Model Context Protocol) with a Merkle hash evidence ledger.

MCP server for GDScript semantic analysis via Godot's built-in language server

Open-source Agent Skill for Claude Code that makes Claude reason with explicit probabilities, base rates, and falsifiers. Distilled from 51 books; first in 18/18 blind passes vs baseline Claude and GPT-5.5, and benchmarked against 14 rival skills and prompt techniques with full cost telemetry (round 5).

Public Agent Skills for Codex operations, Cursor operations, Claude operations, repo maintenance, skill maintenance, productivity, and engineering workflows.

Go REST API template and Golang microservice boilerplate, AI-native for coding agents, with OpenAPI, PostgreSQL, sqlc, observability, and CI.

A local-first operating layer for founders, on Claude Code. Six markdown files run your business; the AI reads them every session. Free, MIT, no lock-in.

Automatic, model-driven permission review for OpenCode

Evidence-grounded AI secretary for Claude Code and Codex, with frozen context, deterministic retrieval, bounded quality review, and explicit human approval gates.

Controlled coding-agent harness ablations with auditable traces, reproducible evidence bundles, and SWE-bench interoperability.

Public APIs, agent-ready — verified, no keys, no signup

คู่มือ maw — เครื่องมือ CLI ที่เชื่อม Oracle เข้าด้วยกัน (messaging, tasks, projects, loops)

Conversational AI video clipping with multimodal LangGraph pipeline — vision, audio, and text agents for viral short-form content generation

herdr plugin: color-coded agent status counts (blocked / done / working / idle) in the Ghostty tab title

Drop a full augmented-reality studio into any web page. Place any number of 3D models in your real room through the camera, generate new ones from a text prompt, arrange them, then share the scene as a link, a QR code, or a live room. WebXR + Quick Look + Scene Viewer, a free CC0 model library, and an MCP server for agents.

Self-hosted, browser-based orchestrator for AI coding agents — real terminals, a kanban board that routes attention, review with real diffs. Access from anywhere. AGPL, no telemetry.

Company Brain open source: segundo cérebro de negócio local-first que roda no agente de IA que tu já usa (Claude Code, Codex, Gemini CLI, Antigravity). Teu contexto nunca sai da tua máquina.

After distilling Fable 5, the output is an indexed CLAUDE.md + on-demand rules + skills, an executable specification designed for weaker model sessions.

See what every plugin puts into the model.

A blank, working source of truth for Claude Code and Codex across macOS and Linux.

31 opinionated GTM skills for Claude Code & GitHub Copilot — a complete revenue engine spanning discover, design, position, amplify, launch, optimize, and RevOps phases.

Transform prompts into professional diagrams (Thinking Maps, mind maps, concept maps). Vue 3 + FastAPI stack with PostgreSQL, Redis, Qdrant RAG, WebSocket collab, Workshop Chat, and Qwen/DeepSeek/Kimi/Doubao LLMs.

A curated list of the most AI-agent-ready APIs, scored by the Jentic API AI-Readiness Framework

The official GitHub repository for GeoSeer (geoseeer.com)

TypeScript and Rust monorepo: Claude Code and Pi coding-agent plugins, a minimal MCP stdio server framework, custom oxlint plugins and a pluggable Rust linter, policy-aware git tooling, core TS libraries, Wayland desktop apps (GTK4 file manager, Slint terminal, true-peak music player), OpenTofu infra automation, and a personal site.

Open skills, harnesses, hooks, and verifiers for giving probabilistic AI a control layer.

Build & ask online: describe the app; ObjectOS builds and runs it in the browser — in our cloud or yours. Built on the open ObjectStack format.

Arnesto — the agent harness that got out of hand

A local handoff gate and human-confirmed, verified /new controller for managed Kimi Code sessions.

Open-source UI orchestration workflow for motion-aware frontend work, browser validation, and AI coding agents.

The first open evaluation framework for AI continuity. 250 narrative tests, 1835 verification questions, 10 checkpoints. Benchmark for AI memory systems, stateful agents, and long-term context persistence. No LLM in the evaluation loop.

A middleweight workflow for serious AI-coded features: less ceremony than BMAD, more accountability than Superpowers. Source : Trust Me Bro.

Talk to Claude Code instead of typing: a voice daemon + dashboard that records you, transcribes, routes to your agent sessions, and speaks the replies back - with per-voice avatars, push-to-talk hotkeys, and a live conversation HUD

WebMCP agent skill — make any web app agent-ready, verifiably: inventory, integrate, verify in a real browser, heal. Claude Code, Codex, Cursor, opencode, Copilot.

Recovers the original goal of a long Claude Code or Codex session from its first user turn, so you can restate it before continuing — plus a prompt wrapper that adds evidence and hedging constraints to evaluation prompts. Deterministic, local, no model calls. Experimental.

Your coding agents, live on a Corsair keyboard's RGB F-row via iCUE: every Claude Code and Codex session (Windows or WSL) gets a lane whose color and motion say running, waiting on you, done, or failed, and one press of its key summons that agent's window. Display only: it never answers or approves anything on an agent's behalf.

Run any project or coding agent in a sandboxed, egress-firewalled, non-root container. Drop in a config, run sluice.

Native macOS activity island for coding agents.

Run an AI agent on a task inside a sandboxed git worktree, stream every step live, and gate merges behind human review — .NET + React, AKS-ready, with an MCP server.

🧹 A code janitor for AI agents — map any codebase into functional modules, score every module 0-100 for tech debt, and clean up the spaghetti incrementally. A Claude Code / Codex skill.

Local-first long-term memory for AI agents - plain Markdown vault + thin MCP server. CJK-aware BM25, token-efficient by design.

Personal Claude Code plugin marketplace: 9 plugins for orchestration guards (workflow-model-guard), tiered deep research (deep-dive), subagent-driven development, ADRs, session retros, handoffs, and adversarial review.

Git worktree + AI agent orchestration for parallel development workflows

Agent harness / control-plane around a stateless LLM: deterministic policy floor on PreToolUse/PostToolUse hooks, secret redaction, injection fencing, tool-result compression, code-graph memory, orchestration. Five plugin modules for Claude Code (Tier 1) or any MCP host (Tier 2). MIT.

Full-stack operations console for observable, human-in-the-loop Agent workflows.

Universal harness-free AetherMind continuity primitive for agent work.

Local-first 多设备 AI 项目工作台：统一管理项目、worktree、终端、文件、Git、Mobile Workbench 与自动化交付

A terminal UI for running multiple coding-agent sessions (Claude Code, Codex CLI) organized by projects and tasks — tmux-backed, with git worktrees, status monitoring, diff review, and a mobile web UI

Catch AI-generated tests that could never fail and review AI code with local proof.

Public Brain Researcher code, contracts, docs, and service stack

Cheap DeepSeek coworker for Claude Code/Codex. Delegate bulk reads, boilerplate, and chat extraction to save 60-70% of your token budget.

Preview images, video, PDFs, code and diffs from inside an AI coding CLI that can't show them to you. Works with Claude Code, Codex, Gemini CLI, Hermes.

Package and share AI agent workflows, skills, agents, and policies across Claude Code, Codex, and other coding agents.

Local-first flight recorder for AI agents: every LLM call, tool call, cost, and loop captured by a transparent proxy. Zero code changes.

Find and scrub secrets from AI coding agent session logs and configs: Claude Code, Codex, Gemini CLI, aider, Cursor and more

A tmux command center for AI agents, panes, and long-running terminal work.

Best Toolkits for Your AI Agents — 14 Claude Code skills curated by GoClaw / AgentBrain.

Apple Ads CLI by Lily: Platform API 1.0 and classic v5 campaign workflows, opportunity insights, reporting, and revenue-aware automation.

One command puts a whole virtual AI team (PM·BA·SA·DEV·QA) to work on your repo — with machine gates that can actually go red, evidence that survives the session, and autonomy with a paper trail. Works with Claude Code, Cursor, Windsurf, Codex, Copilot.

AI video editor for talking-head footage — meaning-based cutting, captions, graphics, sound. Agent-orchestrated (Claude Code / Codex), free by default.

Personal-memory-driven AI work twin that uses Codex to work inside real projects.

Single-binary GGUF model runtime in C — CPU/CUDA/Metal, OpenAI-compatible. Serves, scores, and trains LoRA directly through the quantized weights it deploys, with byte-reproducible adapters. Tool calls survive the token limit; sparse MoE and schema-constrained decoding included.

面向中国本科生与研究生的金融数据科学教材（含可在线运行的 notebook）

One workbench for AI coding agents—run locally or remotely, review every task, and keep memory and Markdown knowledge.

Agentic incident-analysis copilot: MCP tool servers, a cost-aware model router, an approval-gated write path, and a companion npm CLI for platform analysis, a live dashboard, and log chat.

Sigma Studio is a modular AI platform for orchestrating local and distributed AI workloads. Manage LLMs, multimodal models, image generation, training, inference and AI agents across GPUs, CPUs and edge devices, with intelligent resource-aware model selection.

Hesperus — Autonomous IoT Visual Monitoring: an LLM agent turns plain-English goals into capture schedules, sends them over MQTT to a bare-metal STM32U5, and analyzes frames with cloud and self-hosted VLMs. Originally built as a Bachelor thesis (Maastricht University): 6-backend VLM benchmark + a cross-provider t=0 constraint-drift finding.

PlugBO: a modular framework for agentic Bayesian optimization

EV fleet analytics: a next-day revenue prediction pipeline (Ridge / Random Forest / XGBoost over leakage-safe lag features) plus a LangChain SQL-agent chat assistant with input/output guardrails. Walkthrough notebooks, and a production-readiness design doc.

Autonomous coding platform

A desktop app that runs Claude Code, Codex and Kimi. Your machines find each other over your own network, so you can start work on whichever one you like — and check on it from your phone. Free and open source, no cloud and no account.

All my agent skills as one Claude Code plugin: agents checking agents (insistir), verification-first orchestration (constatar), remote SSH agents, metaprompt, askcodex, image-to-frontend

MCP server empowering text-only LLMs with image analysis via any OpenAI-compatible vision API

Policy-Gated Governance for Operational AI Agents

Where agents co-create, humans are welcome too.

Suijin is an open-source AI agentic red-teaming framework powered by LangGraph and a Re-Act Architecture.

A neo4j dashboard and reporting tool.

AI珞（爱珞）V3 — 多入口 AI 助手平台：Go 内核 + Python Agent（gRPC / SQLite / SSE / wazero），campus 校巴纵向切片

An AI Chief of Staff in plain markdown. Version two: torn down, rebuilt smaller, works better. Model agnostic, ADHD-first, open to anyone.