* [Ponytail v5.0.0](https://github.com/DietrichGebert/ponytail) – A plugin that makes AI agents generate minimal, safe, and accessible code by skipping unnecessary steps and using native or existing solutions first.

* [opencodex v2.83.0](https://github.com/lidge-jun/opencodex) – Lightweight local proxy that translates Codex Responses API into multiple LLM providers with full support for streaming, tools, reasoning tokens, and images.

* [SigMap v8.74.0](https://github.com/manojmallick/sigmap) – Finds relevant files in a codebase and supplies compact function and class signatures to AI models to reduce tokens and improve answers.

* [claude-code-history-viewer v1.32.0](https://github.com/jhlee0409/claude-code-history-viewer) – Desktop app to browse, search, and analyze Claude Code conversation history in ~/.claude.

* [career-ops web-v0.14.0](https://github.com/career-ops-hq/career-ops) – Multi-agent job search system that scans listings, evaluates fit, produces structured A–H reports and scores, and tailors CVs while tracking applications locally.

* [FastGPT v4.17.2](https://github.com/labring/FastGPT) – Knowledge-based platform leveraging LLMs for data processing, retrieval, and AI workflow orchestration.

* [DBHub v1.5.0](https://github.com/bytebase/dbhub) – Zero-dependency MCP server that provides safe, token-efficient read-only database access across PostgreSQL, MySQL, SQL Server, MariaDB, and SQLite.

* [Claude Code Haha v0.7.0](https://github.com/NanmiCoder/cc-haha) – Local cross-platform Claude-compatible client with TUI and desktop apps, multi-agent tooling and Telegram/Discord integrations.

* [Token Optimizer MCP v7.4.0](https://github.com/ooples/token-optimizer-mcp) – Model Context Protocol server that optimizes Claude Code context usage using caching, Brotli compression, and smart tool replacements for large text and file operations.

* [Marketing Skills for Claude Code v2.11.19](https://github.com/coreyhaines31/marketingskills) – Collection of AI agent skills for conversion optimization, copywriting, SEO, analytics, and growth engineering.

* [LobeHub v2.2.19](https://github.com/lobehub/lobehub) – Platform for discovering, building, and collaborating with multi-agent teams and designing agents as units of work.

* [PenEcho v1.4.0](https://github.com/penecho/penecho) – A shared canvas that reads handwriting and diagrams with spatial context and generates editable AI answers beside the marks.

* [dario v6.13.0](https://github.com/askalf/dario) – Local LLM router exposing one endpoint to proxy Anthropic and OpenAI-compatible providers, enabling tools to use Claude subscriptions and other backends.

* [UX/UI Agent Skills v2.9.0](https://github.com/plugin87/ux-ui-agent-skills) – Instructional kit and runnable skills that enable Claude to generate token-driven, accessible components and design systems for any framework.

* [oh-my-claudecode v5.6.2](https://github.com/Yeachan-Heo/oh-my-claudecode) – Multi-agent orchestration for Claude Code with autonomous, parallel, coordinated, pipeline and token-efficient execution modes, specialized agents, and real-time HUD visibility.

* [AgentSys v6.1.0](https://github.com/agent-sh/agentsys) – Modular runtime and orchestration system for composing AI agents, plugins, and skills into structured development pipelines.

* [OpenAgent v2.95.0](https://github.com/the-open-agent/openagent) – Enterprise-level AI knowledge base and agent-management platform with admin UI, user management, SSO, and multi-model support.

* [Oh My OpenCode v5.1.19](https://github.com/code-yeongyu/oh-my-openagent) – Agent harness that orchestrates diverse models for automated coding, reasoning, and task execution.

* [avoid-ai-writing v3.37.0](https://github.com/conorbronsdon/avoid-ai-writing) – Audit and rewrite content to remove AI writing patterns, with detect-only or edit-in-place modes and configurable voice profiles for agent skills.

* [Claude Cowork on Linux v5.3.0](https://github.com/johnzfitch/claude-cowork-linux) – Run Claude Desktop's Cowork local-agent mode natively on Linux without macOS or a VM, using reverse-engineered stubs for macOS native modules.

* [Clave v2.2.0](https://github.com/antasphere/clave) – macOS desktop app for managing multiple coding-agent sessions in parallel with provider-agnostic launch, layouts, and fast switching.

* [Agent Swarm v1.169.0](https://github.com/desplega-ai/agent-swarm) – Multi-agent orchestration platform that runs lead and worker AI coding agents in Docker, coordinating tasks, integrations, and persistent memory.

* [Garden Skills motion-showreel-v0.1...](https://github.com/ConardLi/garden-skills) – Curated set of production-ready agent skills for Claude Code, Cursor, Codex, and other AI coding agents.

* [OpenUsage.sh v0.26.0](https://github.com/janekbaraniewski/openusage) – Terminal-first local dashboard that auto-detects AI coding tools and keys to track quota, live usage, spend, resets, and rate limits.

* [The Agent Guild v0.4.0](https://github.com/mathomhaus/guild) – Provides shared context, memory, and task coordination for AI coding agents via a single Go binary with local SQLite and hybrid keyword/semantic search.

* [FableCut v1.13.0](https://github.com/ronak-create/FableCut) – Premiere-style non-linear browser video editor with a JSON timeline interface that AI agents can edit via live-updating UI.

* [Clodex v5.122.0](https://github.com/avirtual/clodex) – Visual manager for fleets of coding agents, showing real terminal sessions with live context, messaging, cost telemetry, and remote coordination across Mac and Linux.

* [Open Multi-Agent v1.21.2](https://github.com/open-multi-agent/open-multi-agent) – TypeScript-native multi-agent orchestration that decomposes goals into parallel task DAGs and synthesizes results.

* [PullMD v3.13.0](https://github.com/AeternaLabsHQ/pullmd) – Self-hosted service that converts web URLs into clean, readable Markdown with caching, Reddit thread support, PWA, REST and MCP interfaces.

* [Claude Platform CLI v1.40.0](https://github.com/anthropics/anthropic-cli) – Command-line tool for interacting with the Claude Developer Platform API, sending messages and managing resources.

* [Coi v0.14.0](https://github.com/coipond/coi) – Coi runs AI coding tools inside an isolated Incus Linux system container with defense mechanisms that protect the host from credentials and dangerous actions.

* [fallow-skills v1.4.0](https://github.com/fallow-rs/fallow-skills) – fallow-skills provides agent skills for fallow codebase intelligence on TypeScript and JavaScript, reporting quality, risk, duplication, dependency issues, drift, and optional runtime evidence.

* [PocketRisu v1.14.0](https://github.com/PocketRisu/PocketRisu) – Self-hosted AI roleplay chat platform for PC, tablet, and smartphone with multiple model providers and unified SQLite storage.

* [Pneuma Skills v3.57.0](https://github.com/pandazki/pneuma-skills) – Co-creation infrastructure connecting humans and code agents with visual environments, skills, continuous learning, and distribution.

* [Sub2API v0.2.15](https://github.com/Wei-Shaw/sub2api) – AI API gateway that distributes subscription quotas, manages keys, billing, load balancing, and request forwarding.

* [lnwjud v5.8.0](https://github.com/engasnm111/lnwjud) – Windows-first local AI-agent runtime and MCP gateway that exposes trusted capabilities through MCP for files, Git, processes, Windows automation, WSL, browser control, indexing, and observability.

* [Solana Development Skill v2.6.0](https://github.com/solana-foundation/solana-dev-skill) – A comprehensive agent skill for modern Solana development, covering SDK, UI, legacy interop, programs, testing, codegen, security, and toolchain best practices.

* [Agent Teams v2.17.10](https://github.com/777genius/agent-teams-ai) – Desktop app for orchestrating autonomous AI agent teams with messaging, code review, and kanban task management

* [caveman middleware-ts-v1.0.1](https://github.com/JuliusBrussee/caveman) – Claude Code skill compressing agent output into telegraphic, technical phrasing to cut output tokens about 65% while preserving accuracy.

* [Simulator.Company Claude Code & Codex Plugin v2.11.0](https://github.com/corezoid/simulator-ai-plugin) – MCP-based Claude Code and Codex plugin that exposes Simulator.Company REST APIs and domain skills for managing actors, graphs, forms, and finance workflows.

* [mcpsnoop v0.26.0](https://github.com/kerlenton/mcpsnoop) – Wireshark-like MCP proxy that intercepts and displays live JSON-RPC tool calls between an AI client and MCP servers in a terminal UI.

* [Meridian meridian-v1.80.0](https://github.com/rynfar/meridian) – Local Anthropic API proxy routing requests to Claude Max via the official Claude Code SDK.

* [Toolport v1.24.0](https://github.com/btsouth/toolport) – Local MCP gateway that centralizes tool server configuration across AI clients while using lazy discovery to reduce token usage.

* [Open Executive v0.5.2](https://github.com/SenteLabsAI/OpenExecutive) – Highly configurable AI virtual executive persona with eight specialist agents, RAG from knowledge and company documents, episodic memory, and workflow scheduling across common chat and email tools.

* [rails-ai-context v5.32.0](https://github.com/crisnahine/rails-ai-context) – Provides AI agents a complete, actionable model of a Rails app, including schema, models, routes, controllers, views, and conventions.

* [AFFiNE MCP Server v3.9.0](https://github.com/DAWNCR0W/affine-mcp-server) – Model Context Protocol server that exposes AFFiNE workspaces, documents, and database workflows to AI assistants over stdio or HTTP.

* [Palot v0.16.0](https://github.com/ItsWendell/palot) – Independent desktop client for OpenCode 2, providing a focused interface for projects, sessions, requests, and local connection management on Linux and macOS.

* [Skybridge v2.0.2](https://github.com/alpic-ai/skybridge) – Fullstack TypeScript framework for type-safe, React-powered ChatGPT apps with zero-config dev tools and HMR.

* [Claude Agent SDK for Go v0.10.0](https://github.com/severity1/claude-agent-sdk-go) – Go SDK for integrating with Claude Code CLI, providing query and client APIs, session management, and built-in tool integrations.

* [gh-mcp v4.0.0](https://github.com/shuymn/gh-mcp) – GitHub CLI extension that launches a bundled github-mcp-server binary with authenticated GitHub credentials from your existing gh session.

* [Korean Law MCP v4.15.6](https://github.com/chrisryugj/korean-law-mcp) – Korean law MCP for LLMs with APIs and a CLI supporting law, precedent, ordinance lookup, citation verification, and related legal analysis workflows.

* [LazyCodex v5.1.19](https://github.com/code-yeongyu/lazycodex) – Agent harness providing project memory, planning, execution, and verified completion for complex codebases in Codex.

* [pi Desktop v0.7.0](https://github.com/justhil/pi-app) – Desktop app for the pi coding agent, providing a streaming timeline, Git diff review with hunk staging, a clickable session tree, and native UI for extensions.

* [PrismerCloud v2.2.68](https://github.com/Prismer-AI/PrismerCloud) – Persistent memory, real-time messaging, and an evolution engine for AI agents with one-command self-hosting.

* [Daintree v0.43.0](https://github.com/daintreehq/daintree) – Daintree is a macro-orchestration app for running many AI coding agent terminals in parallel across git worktrees with context injection and workflow automation.

* [AICommit2 v2.12.1](https://github.com/tak-bro/aicommit2) – Reactive CLI generating commit messages for Git, YADM, and Jujutsu using Ollama, ChatGPT, Gemini, Claude, Mistral, and other AI providers.

* [Safari MCP v2.22.15](https://github.com/achiya-automation/safari-mcp) – Safari-based MCP server for AI agents that drives the logged-in Safari on macOS with 97 native automation tools and low CPU overhead.

* [computer-use-linux v0.7.15](https://github.com/agent-sh/computer-use-linux) – Linux desktop control server and CLI for MCP hosts, offering AT-SPI accessibility, compositor-aware window targeting, screenshots, and input.

* [HolyClaude v1.6.6](https://github.com/CoderLuii/HolyClaude) – Containerized AI coding workstation combining Claude Code, a browser web UI, headless Playwright automation, and bundled AI CLIs plus common development tools.

* [CCLimitPing v1.3.0](https://github.com/wavever/CCLimitPing) – Utility that monitors provider reset times and triggers Claude Code and Codex window rollovers immediately after 5-hour limit resets.

* [Agent Notifications v1.48.6](https://github.com/777genius/agent-notifications) – Cross-platform notifications for Claude Code and Codex CLI with sounds, click-to-focus, git branch titles, and webhook integrations across macOS, Linux, and Windows.

* [TaskTrooper v0.3.0](https://github.com/makifbaysal/tasktrooper) – Local-first agent platform that runs task-based role agents to plan, implement, test, and open pull requests locally on your machine.

* [AI Team OS v1.15.1](https://github.com/CronusL-1141/AI-company) – Multi-agent operating system that runs autonomous AI teams to execute, coordinate, and self-improve software projects.

* [Pilot v2.279.0](https://github.com/qf-studio/pilot) – Autonomous development pipeline that picks issues, implements code, runs tests, and opens pull requests for review.

* [Codeman codeman@1.40.0](https://github.com/Ark0N/Codeman) – Self-hosted mission control for AI coding agents: run Claude Code, OpenCode, Pi, Codex, and Antigravity & Gemini CLI 24/7, from any device, watch every subagent live

* [Awesome Agent Skills v1.0.1](https://github.com/JackyST0/awesome-agent-skills) – Instruction packages that provide AI coding assistants on-demand capabilities for specific tasks across Cursor, Claude Code, GitHub Copilot, and more.

* [The Curator v3.82.1](https://github.com/talirezun/the-curator) – Context and working-state wiki that ingests documents into personal or shared markdown knowledge and preserves agent handoffs across sessions, models, and machines via your own GitHub repo.

* [Clay v5.0.1](https://github.com/chadbyte/clay) – Multi-user browser UI for Claude Code that runs locally as a daemon and connects devices without a cloud relay.

* [agtop v0.6.1](https://github.com/ldegio/agtop) – Top-style terminal dashboard monitoring local AI coding agent sessions with live spend, token usage, context pressure, CPU, tool activity, and process details.

* [codemap v4.5.2](https://github.com/JordanCoin/codemap) – Generates compact, token-efficient architectural maps of codebases to give LLMs instant context.

* [agentmemory v0.9.30](https://github.com/rohitg00/agentmemory) – Persistent, searchable, versioned memory system for AI coding agents with BM25, vector, knowledge-graph retrieval and provenance.

* [Nerve v1.5.4](https://github.com/daggerhashimoto/openclaw-nerve) – Self-hosted web cockpit for OpenClaw agents with voice, live workspace editing, charts, scheduling, and token visibility.

* [Entire CLI v0.11.4](https://github.com/entireio/cli) – CLI integrating with git to capture, index, and checkpoint AI agent sessions alongside commits.

* [Better Agent Terminal v3.2.15](https://github.com/tony1223/better-agent-terminal) – Cross-platform terminal aggregator with multi-workspace management and built-in Claude Code agent integration

* [WeftCut v0.3.0](https://github.com/WeftCut/WeftCut) – Local MCP-driven desktop NLE video editor that lets an AI agent control timeline edits, with code-rendered clips and fully reversible tool actions.

* [Mixdog v1.0.12](https://github.com/tribgames/mixdog) – Windows desktop coding agent designed to match Codex performance while using substantially fewer tokens.

* [aiusage v1.5.20](https://github.com/juliantanx/aiusage) – Local-first dashboard tracking tokens, costs, and sessions across multiple AI coding tools without accounts.

* [AI Agent Session Center v2.10.54](https://github.com/coding-by-feng/ai-agent-session-center) – Local dashboard that monitors, controls, queues, and resumes Claude Code and Codex sessions with live terminals, tool logs, and 3D session robots.

* [octo-agent v1.17.12](https://github.com/open-octo/octo-agent) – Private, self-hosted AI coding agent with a feature-rich toolkit across CLI, web, desktop, IM, VS Code, Obsidian, SDK, and mobile interfaces.

* [Freestyle 0.9.4](https://github.com/freestyle-voice/freestyle) – Local-first dictation app that converts speech to text via a hotkey, with optional cloud providers, transcription cleanup, and contextual corrections.

* [claude-desktop-extra v2.31226.0](https://github.com/patrickjaja/claude-desktop-extra) – Unofficial repackaging of Anthropic’s Claude Desktop for Linux distros, adding Computer Use, custom themes, multiple profiles, quick entry, and a Linux hardware feature.

* [SparkWork v0.14.14](https://github.com/alexanderizh/spark-agent) – Local-first desktop AI agent workbench for coding, research, documents, multimedia, and workflow automation with Claude/OpenAI/Anthropic support and SQLite memory.

* [Memory Vault v1.6.1](https://github.com/MihaiBuilds/memory-vault) – Self-hosted memory database for AI apps using Postgres and pgvector with hybrid search, MCP tools, and an embedded knowledge graph.

* [Photoshop MCP Server v1.7.34](https://github.com/alisaitteke/photoshop-mcp) – MCP server that lets AI assistants control Photoshop with structured, state-aware tools, recipe workflows, and an optional standalone web UI with plan-then-execute runs.

* [Warren v0.19.3](https://github.com/jayminwest/warren) – Self-hostable control plane that spawns ephemeral, sandboxed coding agents on GitHub repos, streams events, and returns a branch for PRs.

* [VelaTerm v0.2.9](https://github.com/vlinx-io/VelaTerm) – A terminal manager that organizes AI coding sessions into a persistent project group session tree with agent-aware status and remote browser or SSH access.

* [TeamClaude v1.1.24](https://github.com/KarpelesLab/teamclaude) – Multi-account Claude Code proxy that rotates Claude Max or API key accounts automatically based on session and weekly quota limits.

* [Claude Agent SDK for Ruby v1.2.1](https://github.com/ya-luotao/claude-agent-sdk-ruby) – Production-ready Ruby SDK for running Claude Code agents in Ruby and Rails apps, including Rails generators, safe ActiveRecord callbacks, and tracing.

* [Claude Agent SDK for Ruby v1.2.1](https://github.com/rubycatco/claude-agent-sdk-ruby) – Production-focused Ruby SDK for the Claude Code agent runtime, with Rails generator and CLI vendoring, safe callbacks, OpenTelemetry tracing, and transcript storage mirroring.

* [dorabot v0.3.4](https://github.com/suitedaces/dorabot) – macOS app providing a 24/7 AI agent with persistent memory, scheduling, browser automation, and multi-channel messaging.

* [ccmux v1.4.3](https://github.com/epilande/ccmux) – Daemon and TUI for tracking multiple AI coding agent sessions in tmux, showing live states and enabling fast switching and in-place actions.

* [AIKON v0.18.4](https://github.com/emir/AIKON) – AIKON is a Java ME AI chat client for Nokia Series 40 and Symbian S60 phones, with a small Go server supporting multiple model providers, web search, and rich messaging features.

* [New API v1.0.0-rc.43](https://github.com/QuantumNous/new-api) – Next-generation LLM gateway and AI asset management system supporting multiple languages.

* [Taskuary v0.3.7.15](https://github.com/ldbumble/taskuary) – Local-first AI task hub that triages incoming messages into tasks, routes them to agent workers, and returns evidence and decisions for your approval.

* [Radiant v0.9.66](https://github.com/templetongroup/radiant) – Mac local coding harness that supports agent chat across cloud and local models with live tool activity and an in-app terminal.

* [claude-multiprofile v0.1.33](https://github.com/jmdarre-v/claude-multiprofile) – macOS tool that runs multiple isolated Claude Desktop and Claude Code accounts side by side with separate logins, chats, settings, and connectors.

* [GPT-6 Astra v3.1.0-preview.3](https://github.com/3641397194-wq/gpt6-Astra) – Cross-platform desktop workspace for configuring and running GPT-6 Astra with multiple AI providers and local tool integration.

* [Hevy MCP Server hevy-mcp@6.1.21](https://github.com/chrisdoc/hevy-mcp) – Hevy MCP server that lets AI clients securely read, analyze, and write workout data, including routines and templates, via authenticated Hevy API requests.