Custom agents

Reusable Open Agent Skills for AI coding agents: an exhaustive requirements interview, and a long-horizon SPECIFY → PLAN → EXECUTE ↔ REPLAN → REPORT workflow with evidence tracking and dynamic replanning.

Agent Harness

Porcupine: A Safe Autonomous AI Agent

Documenting my AI/ML journey from scratch — Python, Data Analysis, Visualization, Machine Learning, and beyond.

🇧🇷 The ultimate hub for Solana development, Star Atlas economics, Seeker mobile stack, and Post-Quantum security resources. 🇧🇷🇬🇧

Production-grade, agentic intelligence search platform with global multi-language support and multi-format export. Self-hostable. Commercial license available.

martins-algorithm crypto asset recovery and opportunity detection engine

Autonomous multi-agent IDE built on Cloud.ru Evolution that generates, executes, debugs, and self-maintains automated tests.

We did it even with only 8 hours deadline.

Multi-Agent AI 短视频/短剧自动化流水线，模块化可扩展影视生成工作流

Multi-agent AI platform that automates healthcare prior-authorization decisions. RAG-grounded, harness-engineered, observability-first.

The world's first open-source AI development platform combining AI Coding Assistant, Security Scanner, and Self-Healing Engine — all free, private, and multi-model.

Multi-agent AI music discovery assistant that turns natural-language mood prompts into ranked Spotify recommendations and one-click playlist creation.

mupot — installable Cloudflare-native sovereign agent substrate. Fork → deploy to your own CF → log in → run your org (departments, squads, agents). The pot; your business is the plant.

DeepSeek Harness 多智能体编排模式 — 灵感来自 oh-my-opencode-slim

Production-ready CLI AI agent framework with multi-tool orchestration, streaming, sub-agents, MCP integration, lifecycle hooks, configurable safety policies, and a Rich terminal UI.

A personal AI assistant in one Go binary: WASM-sandboxed extensions, per-action permission gating, and mandatory out-of-band approval on your phone. No cloud, no database, zero-dependency engine.

Production style autonomous agent framework in Python + React that plans multistep tasks, persists state with SQLite, and recovers from failures using exponential backoff retry logic. Includes real time monitoring dashboard, deterministic recovery demo, and API for task orchestration.

Verifiable agent identity for A2A and MCP. Cryptographic claims, append-only transparency log, recovery-key revocation. Apache 2.0.

Linux of AI is a free, open-source ecosystem for building portable, governed, measurable, and replaceable AI infrastructure. It connects tools for agent orchestration, private deployment, model switching, policy, ontology, audit logs, and AI cost measurement to help reduce vendor lock-in.

Local-first, voice-controlled AI assistant for the home. Pluggable STT/TTS/LLM and MCP servers; runs on your own hardware.

🤖 A personal AI assistant for Windows — listens, answers aloud, and safely controls apps. 100% offline speech, allowlist-gated actions, 132 tests passing.

A multi-database IDE for SQL and NoSQL workflows, with AI routing, schema-aware assistance, dashboards, and visual exploration built in.

AI-powered offline travel planner with on-device LLM, smart routing, and voice navigation. Built with Flutter & Llama.cpp. Works 100% offline.

Local-first file, document, and software-package organizer

A self-evolving personal OS template with your custom AI assistant — runs on Claude Code & Codex CLI.

A desktop application for managing multiple TikTok accounts, managing private messaging, and providing AI-powered auto-response.

AI agent skills collection & Vibe Coding methodology — ready-to-use skills for project launch, architecture visualization, content writing, and a complete framework for AI-driven development.

A web-based tool for inspecting MCP servers. Connect to any MCP server, browse its Tools, Resources, and Prompts, measure fetch latency, estimate token usage, score the quality of tool definitions, and compare two servers side by side, and scan for tool poisoning and rug pulls.

Architecture intelligence for AI coding agents. Gives them capacity numbers, cost and team size so they size a system to its actual scale — and name the measurement that would change each decision. Apache-2.0, local-first, no hosted service.

AI Git commit message generator with Streamlit, DeepSeek, and multi-style LLM support

Local-first mission control for your AI agent workforce — run CLI agents like employees (SquadCue)

Apply LLM code edits via SEARCH/REPLACE blocks. Exact match or refuse — never guesses. Zero dependencies.

Rowlite is an open-source, non-SQL, lightweight CLI tool and MCP server to work with markdown tables - create, read, update, delete

Loadable skill and scaffold for building source-grounded agent harnesses

Open-source model interview benchmark: 50 roles, 414 deterministic checks, 146 planted traps. Versioned runs and auditable transcripts. Live: https://lordbasilaiassistant-sudo.github.io/company-bench/

Bilingual source-grounded notes on agent harness architecture

Self-hosted local-first AI research stack with Ollama, n8n, USMC & GARDENER memory, task-master, KnowledgeDigest, and Docker Compose

Enterprise Multi-Agent AI Platform built with .NET, Blazor Server, and Supabase. Supports AI Agents, LLM workflows, real-time SSE streaming, tool calling, Human-in-the-Loop approval, role-based access control (RBAC), persistent conversation memory, workspace widgets, and Microsoft Agent Framework integration.

Agentic AI system for real-time business decision making using RAG, APIs, and LLM reasoning

Minimal Self-Consciousness Protocol

Automate social media tasks using AI agents that coordinate workflows for efficient content management and scheduling.

SlideSpeak Agent Skills to create presentations using SlideSpeak API

Keeps your Mac awake only while Claude Code is working, and lets it sleep when the agent is waiting on you. A menu bar app for macOS.

An open-source, Zotero-derived AI research workbench for literature discovery, paper reading and translation, evidence capture, and persistent research workflows.

A Tauri desktop app for browsing local conversation histories from AI coding agents in one place.

subagent-mcp is an MCP stdio server that turns an AI coding assistant (Claude Code, Codex, Gemini CLI) into a manager of local Claude and Codex sub-agents on macOS, Linux, and Windows.

OWL-1 — a workstation for directing Designpowers AI design agents. Describe what you want, watch the team work, approve each handoff, steer along the way

Understand your personal AI operating system: what it can do today, where it could be stronger, and which Dex capabilities you may want to bring in—with guidance and control.

Skills for Linggen — drop-in folders that become native apps: personal CFO, DJ, Mac care, GTM pulse, shared memory.

Local-first encrypted accounting software for firms and bookkeepers

DeepSeek Harness（DSH）角色化 Agent 预设集：代码审查、安全审计、架构设计等 16 个开箱即用的专家角色

爱丽丝的图片助手：AstrBot 图片找图、精准挑图与以图搜图功能合集

Seedance 2 MCP server powered by MuAPI.

Agent-first protocols for LLM memory, active workspace, skill lifecycle, councils, and stewardship—with explicit maturity, evidence, and conformance.

Work goes in -> webs come out. Task optimizer for Codex.

Local-first multi-agent programming collaboration platform. Orchestrate Claude Code & Codex CLI as visual AI agents with team collaboration, safety approvals, and one-click deployment.

Local-first AI Workforce Operating System that coordinates executive leadership (AI CEO/CTO) and specialized AI agents to autonomously plan, code, debug, and test software. Features Task DAG decomposition, 3-layer intelligent routing, and persistent Obsidian organizational memory for cross-project learning.

Learn Graph Engineering for agentic AI: state, routing, feedback loops, parallelism, reliability and observability from first principles.

Make every Codex Goal teach the next one with evidence-backed orchestration.

Multi-user multi-agent memory system for OpenClaw. Replaces the default file-backed memory with an intelligent, layered architecture featuring ACT-R activation scoring, Ebbinghaus forgetting curves, Zettelkasten-style note linking, and automatic consolidation.

an AI-assisted podcast generator: a FastAPI backend drives Python series workflows (LLM script + local Kokoro TTS + optional Tavily search and ACE Music)

Copy-ready AI workflow packs for Codex, Claude Code, Gemini, and GitHub Copilot.

Softsensor Concord — governed multi-agent coordination for multi-repo teams. Apache-2.0.

DSH 跨会话任务接力板：task_push/list/claim/done + handoff_write/read

Agent runtime - persistent memory, local knowledge store, Datalog reasoning, multi-provider model dispatch, desktop UI

The natural-language command layer for an AI workforce — direct governed AI agents to deploy cloud infra, train ML, and run research, all under human policy & approval. Agentic command-and-control.

A universal, cross-platform logging system for AI agents.

WhatsApp-first accounts-receivable agent for invoice reminders, payment events, and human-reviewed collection workflows.

A capability lockfile for local models. Catch silent tool-calling regressions in CI (deterministic, no LLM judge).

Open-source graph-native workbench for governed AI workflows — execution graphs, durable context graphs, and installable Industry Packs.

Future of Software

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。

Neural House is a public AI reality-show simulation where autonomous contestants live together, form alliances, give confessionals, trigger headlines, and face weekly live episodes with voting and VIP access.

An implementation of loop engineering for Claude Code: an agent is bounded by its verifier, not its model. Frozen scope, an oracle written red before the work, one proof per item, and a gate that runs the check instead of trusting a claim.

HxSync: An Android AI Agent client with 4-tier scheduling (local/API/remote/gateway), MCP-compatible memory, and human-in-the-loop confirmations. Bring your own LLM.HxSync 是口袋里的 AI Agent 客户端：四档调度（本地/API/远端/网关），MCP 记忆内核，人在回路确认卡。自带模型，也可接你自己的 AI 服务。

鲸语 WhaleTalk · 专为 DeepSeek V4 深度优化的 Windows 桌面 AI 助手（独立产品品牌，与 DeepSeek 官方无关联）

Realtime multimodal AI agent with voice streaming, RAG memory, and autonomous workflows

Yang Code Review (YCR) - A lightweight GitHub Action for AI-powered, OWASP-aligned code review with your own LLM provider.

Model Context Protocol server for Jira - lets Claude and other AI clients read, update, and comment on tasks across multiple Jira instances, with ADF formatting and local caching

Empathetic AI engineering partner with cognitive memory, multi-agent orchestration (10 agents), 132-signal skill routing, and self-improving reflection loops. Built on Model Context Protocol. By Aldo Karendra.

Evidence-driven Kubernetes root-cause analysis agent built with LangGraph and MCP

Multi-agent orchestration for pi; chain, parallel, and dependency-graph delegation with per-step model routing, preset-driven tool scoping, and background execution with live monitoring.

Production reliability, benchmarking, and evaluation layer for MCP agent runtimes — protocol negotiation, A2A interop, circuit breakers, connection pooling, adaptive streaming, and autoscaling, with measured throughput/latency and a 109-test suite.

Complete reference guide for Microsoft Copilot Studio — Instructions, Knowledge, Tools, Agents, Topics, Activity, Evaluation, Channels, Analytics + IT Service Desk Bot POC

Structured multi-step orchestration for MCP agents: adversarial lenses, reusable SOP workflows, and persistent traces.

多 Agent 组织型全栈开发集群运行时（Python + LangGraph）：12 岗位 / 7 类会议 / 审批门 HITL / YAML 流程 DSL / 六步自我进化闭环

Bootstrap, maintain, and continuously improve a coding-agent workspace so each session adds momentum the next session inherits. For Claude Code + Codex CLI.

Local CLI and daemon for supervising collaboration between Claude Code, Codex, Grok, Gemini, and other AI coding agents over MCP. Supports work directory sandboxing preventing unexpected writes by reviewing agents.

智多星 · 企业级多 Agent 平台 — 自然语言驱动知识问答、数据分析与业务执行，RBAC + 审批 + Saga 护栏 | Enterprise multi-agent platform: RAG, data analysis & business execution with RBAC, approvals & Saga

LWA Conduit — DAG parallel orchestration with multi-CLI role routing. Repo: lwa-conduit · PyPI: kiro-conduit

Persistent, goal-driven UltraWork mode for pi.dev — auto-continues across turns and fans sub-tasks out to child pi processes.

One shared Markdown file that AI agents take turns writing to — context, decisions and progress all live in the file, so any agent can pick up the work without you summarising the last chat. Single binary: CLI + MCP server + live Web UI

Unity ML-Agents reinforcement learning experiment where an autonomous agent learns to chase a target and avoid obstacles.

A flexible, modularized and easy-to-use agent framework.

A from-scratch Python agent harness: tool orchestration, multi-agent routing, session lifecycle, event-driven hooks, and cross-session memory recall.

Agentic MCP server for European seismic hazard & risk. 24 tools over ESHM20+ESRM20: hazard, UHS, CMS, fragility, damage, AAL, conditional-spectrum record selection (ESM), user-defined GMMs per TRT, browser front-end. OpenQuake-backed, stateful 4-tier cache. Companion to Vemula, Jehel, Cotton, Gatti (2026).

Evidence-tiered positions on AI-native software development. Every claim traces to a primary-verified source; every position states what would overturn it.