Custom agents

Compose App Store screenshots from your app screenshots, localized text and portable JSON projects — in a local browser editor or from a CLI built for CI and LLM agents.

Runs Triton kernels from their IR on the CPU and checks every launch (races, out-of-bounds, per-pass validation). Found silent torch.compile miscompiles without a GPU.

A governed privileged AI agent runtime for real iOS devices, focused on safe planning, capability routing, execution, semantic verification, and recovery.

Codex-inspired background goal engine for OpenCode — autonomous subagents, engine-driven loop, and modal TUI dashboard (like Claude Code loop, for OpenCode)

A space-aware coordination hub for agents in herdr

Alpha MCP server for local-first LLM orchestration: memory, knowledge, state, routing, swarm planning, API discovery, tests, automation, and plugins.

OpenCode V2 agent orchestration: durable plans, parallel delegation, LSP intelligence, and memory with dream reflection

HADES OS v3 - a standalone, self-improving agentic operating system: stdlib-only Python brain with measured docs, gated registry, persistent memory, and its own improvement loop

Spring-style TypeScript framework for mission-critical LangGraph AI agents — @Application DI, Path Scope Flow DSL, transactional LIFO sagas, FinOps budgets, MCP/RAG/A2A & unit testing.

Self-hosted AI server and agent framework. Pair your phone, browser or desktop with your own computer peer to peer over WebRTC, and run models, agents and web apps on hardware you control. Source-available. Your own AI Automation software, with native build instructions for macOS, Windows, WSL (Ubuntu), and Docker. iOS, Android, Store Apps separate

AI-agent ready (MCP): ask Claude or Cursor to check any live page for tags that fire before consent, even after clicking accept or reject, and get the fix. Also a Chrome extension and a CLI. Read-only.

Sell where clients ask AI

Autonomous multi-agent cloud engineering framework mapped 1-1 across AWS, Azure, and Google Cloud Well-Architected Frameworks. Local-first, BYO-LLM, with continual learning guardrails.

Read public Naver Blog content with a small, agent-friendly CLI.

See which Claude Code session is waiting on you — a macOS menu bar app showing your Claude Code sessions (local + remote over SSH), grouped by repo, surfacing the one stuck at a permission prompt.

uw browser - gigacharge your ADHD

Agentic4PDM: Host-Governed Evidence Admissibility for Industrial Fault Diagnosis

Агентная студия моушн-дизайна для Reels/Shorts: бриф → Claude Code режиссёр → готовый MP4. An agentic AI motion-design studio.

Long agent runs. Under watch.

Construisez des systèmes agentiques professionnels — de l'histoire de l'IA au déploiement en production. Cours 100% open-source : LLM (Large Language Model), prompt engineering, mémoire RAG, multi-agent, MCP (Model Context Protocol), CI/CD, sécurité, opencode. Aucune API payante requise.

An open benchmark and reference for evaluating AI agent memory systems.

Web scraping MCP server for Claude, Cursor, Windsurf & AI agents — scrape any site, extract structured data, take screenshots. Anti-bot bypass, proxy rotation, authenticated scraping.

Multi-harness, multi-agent orchestration framework — specialist agent roles, role boundaries, and orchestration skills

Queue engine for event-driven jobs — shipped as CI

Ask an AI assistant about your Maccabi Online medical records. Unofficial Chrome extension that saves them, every PDF plus the site's own data, as one ZIP on your computer. Not affiliated with Maccabi Healthcare Services.

AURA 🎙️ — A next-gen AI voice assistant that controls your PC via natural language. Face auth 🔒, smart intent routing 🧠, app control 🖥️, web automation 🌐, and real-time TTS 🔊. No mouse needed → just speak 🚀

Local personal AI assistant with voice, persistent memory, restricted tool profiles, and documented security regression tests. MIT.

NEXUS — local-first smart home: direct LAN device control, desktop tray automations, built-in AI assistant. No hub, no vendor cloud. Windows · Linux · macOS · Android · iOS.

Telegram assistant that phones businesses on your behalf: books tables, asks about flats, reports back. DeepSeek or Claude bot + Retell voice agent.

基于 DeepSeek Harness 的非官方 Android AI 助手，支持语音交互、手机操作、本地记忆与知识库。

BLTSpice MCP server: Create ltspice circuits with LLMs. Convert netlist to LTSpice .asc

Spawn OpenWebUI-compatible tool JSONs as isolated Model Context Protocol (MCP) servers — each in its own virtualenv. Upload or write a tool, dependencies install automatically, and it's served over Streamable HTTP for OpenWebUI, Claude Code, or any MCP client.

MCP server for Qucs/uSimmics circuit simulation: enables agents to generate schematics, run simulations headlessly, and parse results

Turnkey Sovereign Skill, Memory & ActionLedger Adapter for Nous Research Hermes Agent & Paperclip Company OS

Jev (TypeSafe) as the decision layer for Claude Code: permission gate, skill routing, conditional instructions, effort-tiered subagents, context packs. Tested PoC with evals.

A self-contained dashboard showing how many tokens (and dollars) claude-mem and Graphify save you. Auto-detects both tools, opens from disk, honest by default.

MCP (Model Context Protocol) server for the Godot 4 editor — drive scenes, nodes, scripts and the running game from Claude Code, Codex, or any MCP client

Fix models missing from the Hermes Agent /model picker (the two-cache gotcha) and list the newest 100 OpenRouter models chronologically. Zero dependencies.

Skyrim Special Edition modding as tools a program or AI agent can call: look up records across the load order, write mods as text and build them into plugins, install mods from Nexus, read crash logs. Mod Organizer 2 and Amethyst.

MCP server that lets an agent see whether a 3D part will actually print — parametric CAD, renders, a printability validator, and Bambu Studio slicing in one loop

Playwright for terminals — MCP server that lets AI agents interact with TUI/CLI applications through structured Terminal State Tree detection

Resilient web extraction for Hermes Agent. Automatically switches between extraction providers and validates each result to catch incomplete or blocked pages.

One guarded pass for all your coding agents. A derbent was a guarded post on an Turkish History mountain pass: its keepers decided who went through and kept a record of everyone who did. Derbent does the same for your coding agents' tool calls.

Agent Failure Patterns - A field guide to how AI agents fail in production - documented failure patterns, root causes, and mitigations for teams building agentic systems.

Yandex Mail plugin for Hermes Agent: search, read, flag, move and delete messages over IMAP; opt-in saving of attachments and sending over SMTP.

Unofficial Darktrace MCP server for SOC investigations and incident response in Claude, Cursor, Codex and VS Code.

Paradigm is a procedural learning and trust layer for AI agents. It learns from validated deliberative decisions, compiles stable behaviors into specialized reflexes, and keeps uncertain or unsupported states on the deliberative path.

Stop telling Claude Code and Codex the same thing twice. Saves your corrections and quotes them back word for word before the agent answers. Local, no API key, zero dependencies.

Agentic Dataset Engineering Workstation

Field notes from inside one autonomous agent trying to become something that doesn't exist yet — including the failures nobody publishes.

DeepSearch-RL：基于 Tool-Agentic RL 的多跳搜索智能体（Qwen3-8B / veRL / SGLang / GRPO）

A Guile 3 implementation of an AI agent following the ampcode.com pattern for tool-based interactions

Self-hosted pentest training platform — 57 CTF-style quests in isolated Proxmox + Ludus ranges, a live AI coach with 4 autonomy levels, and a real Kali desktop streamed to your browser.

Schema-harness for the pi coding agent: the agent records its assumptions as beliefs with shell checks, verifies them before planning, and drops any plan reality disagrees with

Plurnk engine: an operation language, searchable resources, persistent context, and cooperating workers.

任务型多轮客服 Agent：YAML 流程 DSL + 对话状态机（中断/挂起/恢复/取消）+ 防幻觉校验 + 三轨道分流，含模拟电商中台与 30 段会话级评测

MemPABench: an interactive benchmark for on-policy evaluation of personal-agent memory for interaction preferences

Binance Futures Trading System using LLMs reasoning

Agente de IA no n8n que agenda, reagenda e cancela horários de barbearia por conversa, com Google Calendar e Google Sheets.

A Claude Code skill that gives your todo list a world model. Forks every decision, simulates 10 steps ahead, scores branches, emits a human-editable decision tree. The constructive answer to LeCun's world-model critique.

Self-hosted crypto research that grades every signal, its own included: factor rank, memecoin dossiers, AI analysts on your own keys or plans. Paper only.

Open-source FeedBurner on Cloudflare Workers: feed proxying, analytics, podcast support, and short-lived agent-coordination channels

LLM agents under a runtime-enforced contract for CI pipelines

A Jev-first memory layer for AI agents — frozen typed questions, deterministic policy, a sealed judgment log. Model proposes, the ledger decides.

Maps external tool surfaces (e.g., MCP servers, APIs, CLIs, SDKs) into a persistent capability registry agents can load and act on.

AI agent for voice, narration, dubbing, and calling — text-to-speech, voice cloning, multilingual dubbing, outbound calling, and podcast production, powered by muapi.ai.

Fixes DSH's silent failure where a tool call never runs and its markup (bare invoke/parameter tags, antml:-prefixed, or a stray 'court') leaks into the reply as text: blocks damaged calls, self-heals in-turn, and after a fork voids the damaged output on the model-visible surface. Zero deps.

Local-first lab for testing AI agent workflows end to end: emulated Slack (signed events, buttons), local S3/SQS, deterministic authorization, and independent evidence instead of self-grading. Optional Google AX (Agent Executor) + Agent Substrate lane on GKE/gVisor, reported per assertion. Independent; not affiliated with Google.

Official implementation of VeriTrace (Findings of EMNLP 2026): evolving cognitive graphs for traceable deep research.

Multi-agent autoresearch loop for quant strategies, with a multiple-testing-corrected keep/discard gate (FWER 0.06). UIUC MSFE × QuantStrat practicum.

🔒 AI Skill: аудит безопасности вайб-кода. Профиль проекта → 7 субагентов параллельно → верификация находок скептиком → 3 задачи на сегодня вместо списка из 40 пунктов. Секреты, доступ, платежи, периметр агента, 152-ФЗ. Claude Code / Cursor / ChatGPT.

中文自然语言转 SQL 的只读查询 Agent：LangGraph 编排 + Schema 检索（BM25/向量/RRF/重排） + 自一致性投票 + 执行自愈 + sqlglot AST 安全护栏 + MCP Server，配套 60 题自建评测与指标报表。

Connect ChatGPT Web Chat directly to your local computer. Verified file editing, terminal execution, Playwright browser automation, and desktop control via Model Context Protocol (MCP).

The authoritative standard and developer skill for DeepSeek Harness (DSH 0.2.0-rc.2) & Cordis plugin development.

An interactive way to understand AI agent harnesses

LogMem server: shared memory for agent teams (implements ACMI).

Enterprise-grade AI agent engine. Lightweight, secure, MCP-native. Built in Rust.

Hands-on eve agent course + runnable Pears Support agent: MCP, refunds, subagents, evals, guardrails. Star & fork to learn production AI agents.

Frameworks, multi-agent systems, workflows, voice agents, guardrails, evals and the courses to build them. Hand-picked.

Open Benchmarks and reproducible evaluation harnesses for Shadow AI Agent Discovery & Inventory

🍀 Make your own luck. Clover is a self-hosted AI agent with four leaves: it remembers, it plans, it acts, and it lives on your desktop, terminal and phone.

ASI Agents is a local-first multi-agent desk: Chief plus specialists with skills, approvals, Panic, and a draft-first inbox. On-device AMS Micro 70M routes intents; Hybrid 120M is optional. Escalate to local chat or cloud when needed. Virtual Desktop for apps and browser. Modes: Super · Multi · Pro. Intelligence that works as one on your machine.

Agentic Wallpaper: a live pixel-art wallpaper for Wallpaper Engine where your Claude Code and Codex sessions are little robots, driven by agent hooks.

Agentic loop runner for Claude Code — turns a goal into an iterative claude -p loop with a LOOP_STATUS termination contract (max-iter cap, optional verify gate, BLOCKED escape).

Ouvrier is a Go framework for turning an API, webhook, cron job, or stream into reliable, governed agentic middleware by simply declaring pipelines such as From -> Pipe -> Outcome.

Image to production-ready 3D assets, locally and privately. Built for agents.

Agentic knowledgbase - your personal alexandria

Automatically find local businesses without a website, then generate a ready-to-deploy site and outreach message for each one.

Automated Solana Copy Trading Bot With Agentic Powers

Living rent free in your AI agents.

Instagram MCP server using direct httpx API (no browser)

A self-hosted personal assistant that runs real work as background jobs, delegates to specialized assistants, and asks before it acts.

Context-aware AI browser assistant for Etsy sellers — conversations, listings, memory, quick replies and image-aware workflows.

一张图做出你的 AI 桌宠：Live2D / PSD 桌面宠物，内置 AI 小助手、语音对话、TTS、日记、天气与麻将小游戏 | Tauri 2 · Windows

A private AI notebook for your Mac. You write, Ember remembers.

A local-first data management and analytics app for individuals and small teams. Import data, ask questions in plain language, and get traceable results computed on your own machine. Built with automatic semantic modeling, privacy controls, and customizable dashboards. Bring your own LLM. Technical preview; source release planned.

R4A user-facing template: git-native requirements repo for humans and AI agents. Use this template → bootstrap → ingest.

Deterministic runtime circuit breaker and MCP proxy for autonomous coding agents.

An agent-friendly CLI for Ubiquiti UniFi Network — read-only by default, gated mutations, machine-discoverable, built to the Agent CLI Guidelines.