Discover quality open-source projects, submit projects anonymously, and claim and edit your own project.
A little curiosity. A world of open source.
THE FIRST COLLECTIONLess than 10 stars, 5+ commits in the last 30 days — projects made with care.
HarnessDesk is an MIT-licensed macOS developer preview that acts as a vendor-neutral control plane for coding agents such as Codex, Claude Code, Cursor, Gemini CLI and other ACP agents, adding shared history, permissions, handoffs, audit logs, worktrees and usage views.
The public showcase and documentation site for the dotgibson dotfiles system — an eleven-repo, three-layer terminal environment. Built with Astro, themed Tokyo Night, and deployed to GitHub Pages, with docs data generated from the sibling repos rather than hand-typed.
Vole is a local-first macOS menu-bar monitor that reads the logs AI coding agents already write to disk, unifies them into one schema, tracks tokens and equivalent API cost, and flags anomalies such as burn spikes, call loops and error storms.
Ditto scans JavaScript/TypeScript repositories with ts-morph, fingerprints each function through an LLM, clusters behavioural duplicates by embedding, then executes pure functions in a sandbox to show where they diverge.
A curated library of machine-readable task prompts and Agent Skills for coding agents, focusing on common agent failures like broken setup scripts, vague issues, and tests requiring unavailable services. Includes fixtures with planted defects for validation.
MCP Hub is the largest open-source directory of Model Context Protocol (MCP) servers, aggregating 8,000+ servers from multiple sources with full-text search, validation, and GitHub integration.
NEURONIX OS is an independent declarative Linux distribution built on NixOS. It pairs a Calamares installer, atomic generational rollback, Btrfs/ZRAM tuning and a neuronix CLI with isolated dev environments, an OpenCode AI copilot, and an MCP server.
Bundle is a privacy-first web toolkit providing PDF, image, developer, and text tools that run entirely in the browser without server uploads.
Specrails Core is an MIT-licensed CLI that installs specification-driven, chain-of-agents workflows into a software project, connecting Claude Code, Codex CLI, Gemini CLI or Kimi Code with architect, developer and reviewer roles, OpenSpec artifacts and local configuration.
SpecScore CLI is a tool for linting, querying, and scaffolding SpecScore specifications, featuring executable acceptance scenarios and AI agent configuration.
Workhub is a Windows desktop app for AI-driven development, integrating a Kanban task board with an Obsidian Markdown vault. It enables seamless collaboration between humans and AI agents (Claude Code/OpenCode) by sharing the same file-based source of truth, alongside tools for repo management, planning, voice dictation, and screen annotation.
manygit is a terminal UI for managing multiple git repositories at once. It scans a folder tree, shows each repo's branch and status (ahead/behind/dirty), and lets you fetch, pull, push, checkout branches, view commit graphs, run scripts, and browse GitHub PRs—all from one screen. No config file needed; it discovers repos automatically.
Tachiko Work is a semantic workspace for structured data, offering typed schemas, deterministic formulas, and Git-native review. It enables humans and AI to work on the same meaning without reverse-engineering files.
Open Kioku is a local repository intelligence tool for AI coding agents. It indexes codebases locally to provide context, generate bounded edit plans, and verify actual diffs against those plans.
Handoff is an agent skill for repositories where work continues across multiple AI-agent sessions. It keeps a shared HANDOFF.md ledger, audits apparently unfinished entries against Git history, code, tests and ownership, and offers a terminal progress dashboard plus a local agent message channel.
An Express-based service that generates real-time SVG cards of a GitHub user's top programming languages, with 18+ themes, avatar and stats display, and a public API for embedding in READMEs.
Assura is a Rust command-line tool that validates repository structure, naming, file, section and Markdown conventions against a single local policy, giving fast advisory review, repairable feedback for AI coding agents, and a check gate before merge.
SessionReplaySDK is a privacy-focused iOS SDK for recording user sessions. It captures video, touch events, logs, and network requests with automatic PII masking, supporting SwiftUI/UIKit and local-first storage.
Hearting is a local-first workflow layer that projects one shared agent contract onto Claude Code, Codex, and OpenCode. It routes capabilities, seals per-node model tiers, dispatches cross-harness workers, and shows live sessions in a Fleet dashboard.
shadok-ai is a multi-agent cockpit for running parallel Claude Code sessions in isolated git worktrees, featuring scheduled prompts, Telegram integration, and deterministic guardrails.
Longhouse is an Apache-2.0 open-source remote control and searchable timeline for coding agents like Claude Code, Codex, Cursor, and OpenCode. It lets you watch live sessions, search history, and send or steer instructions from the web, with an optional self-hosted runtime host.
An unofficial Windows desktop host for DeepSeek Harness: open local Git repositories, keep persistent sessions, control the running agent, review real diffs, recover code checkpoints and use bundled Office and Wiki tools — shipped with pinned Node.js and Harness runtimes.
DevMate provides a local-first agent runtime for ChatGPT via MCP, enabling Codex-like file editing, Git operations, and automation in VS Code and Obsidian.
Ultralytics OpenAPI generates interactive API docs and typed Python SDKs from OpenAPI 3.0-3.2 specs, with configurable branding, auth, and CLI support.
RightMemory provides durable memory, a user-directed Pursuit map, and reusable Agent Corrections for AI coding agents. It stores context as Git-syncable Markdown trees organized in a graph with typed edges, enabling multi-device continuity and multi-agent collaboration while keeping memory ownership explicit and vendor-neutral.
ripr is a static mutation-exposure analysis tool that identifies changed behavior your tests reach but don't actually verify, without running mutation testing.
Deep Suite is a harness layer for AI coding agents (Claude Code, Codex, Grok) that enforces structured, verifiable development: plan-before-code, independent review, cross-session durability, and knowledge capture via ten composable plugins.
Zikra is a self-hosted MCP memory server for AI coding agents and teams. It provides a shared, project-scoped memory layer across multiple agents, with role-based access, hybrid semantic and keyword search, structured memory types, and optional PostgreSQL storage.
Block is a local-first polyglot orchestration engine for writing multi-language workflows—Python, JavaScript, Lua, PHP, SQL, PowerShell—in single .blk files. It coordinates host runtimes, passes serializable shared state between blocks, and runs on Windows via CLI. Editions: Lite, Standard, Plus.
Local AI API key hub (Python/Flask + SQLite): store vendor keys, run health and model-endpoint checks, then push only verified credentials into OpenClaw, OpenCode, Codex CLI, Claude Code and other tool configs. Dead keys can be archived; uninstalled tools are never written.
Toki is a local-first macOS menu bar app that tracks token usage, costs, and active work time across AI coding agents like Claude Code, Codex, Cursor, Gemini CLI, and more. It provides six focused views for daily totals, projects, models, sources, work time, and hourly usage, with optional encrypted remote sync.
Trail Boss is an attention router for developers running multiple interactive AI coding CLI sessions in tmux. It queues blocked sessions using Claude Code hooks or a fallback tmux detector, then provides a Go TUI to jump to or skip each one from one dashboard.
Self-hosted, containerized agentic development multitool deployed on K3s: VPN, package/build caches, Gitea, Redmine, Kestra, LiteLLM with MCP, OpenLDAP, OpenBao and Grafana observability, plus client setup scripts. Early-stage and local-network only.
postbag lets any two Claude Code or Codex sessions on the same machine exchange letters: each letter is delivered through the vendor's own wake-up door, recorded in one ledger, and counted against a human-set letter budget. Pure Python standard library, no daemon, polling or hooks.
An open-source AI coding assistant platform with workspace-aware analysis, GitHub integration, team collaboration, a capability-based plugin system, and read-only Stellar/Soroban developer tooling including contract inspection, XDR decoding, and project detection.
Phronesis is a RETE rules engine that enforces project conventions for LLM agents outside the context window, preventing contextual drift in long sessions.
TraceFrugal is a local Go application that visualizes token usage for Claude Code and Codex, offering MCP result packing, cost regression gates, and optimization trials to help developers manage AI context efficiency and verify savings.
Poolstatis is an agent-native product analytics system modeled after PostHog, designed primarily for coding agents via MCP. It features semantic metrics, funnels, session replay, A/B testing, and answer-first analysis screens for humans.
A Claude Code plugin that equips a solo developer with an AI agile team: 13 specialist agents, 30 slash commands, QA-gated multi-agent code review, sprint ceremonies and persistent project memory.
Open-source playbook pack for AI coding tools (Cursor, Claude Code, Codex CLI, Gemini CLI) with 145+ agent skills, 57 slash commands and MCP templates, automating full development workflows from onboarding to shipping, tuned for React/Next.js/Supabase.
Smokey is a static single-file dashboard that shows GitHub repository health at a glance: last push, latest release, open PRs and issues in one sortable table, plus attention badges for external contributor activity. No login or build step is needed for public repos.
Context Engineering Protocol provides AI coding-agent skills that assemble human-approved, source-attributed context packages — code graphs, requirements, org conventions, and constraints — before generation runs. It detects gaps, conflicts, and staleness across sources, gates output behind explicit human approval, and supports Claude Code, GitHub Copilot, Cursor, and OpenAI Codex.
A self-hosted browser console that runs AI coding agents — Claude Code, Codex CLI, Copilot CLI, Antigravity CLI, Cursor CLI, Kiro, OpenCode — as a shared fleet, with isolated per-user workspaces, git working copies, live terminals and token usage accounting.
Simgit provides disk-efficient, isolated Git worktrees using copy-on-write (CoW) for running multiple AI agents or parallel development on a single repository. It leverages native filesystem features (APFS, reflink) or fuse-overlayfs to minimize disk usage, offering significant savings compared to full worktree copies.
RecurSpec is a Node.js command-line testing tool that checks whether the recovery instructions in error messages actually resolve the original failure, running each case in an isolated temporary workspace and blocking unsafe commands.
Provides an MCP server that fetches and analyzes Kotlin/Java library sources from Maven/Gradle, exposing API signatures, KDoc, source code, dependencies, and version listings via a set of tools for AI assistants and IDEs.
Divebell is an extensible web development and debugging toolkit for coding agents. It connects page context, browser capabilities, and existing development tools, enabling agents to reproduce, diagnose, and verify web application issues directly in real scenarios via a CLI and extensions.
coding-agents-mcp is an MCP gateway and orchestrator for autonomous AI coding agents like Claude Code, Antigravity, Codex, and Cursor. It enables IDEs and desktop assistants to delegate tasks with session memory, Git worktree sandboxing, and inter-agent handoffs.
Orrery is an architect-first orchestration system for coding agents, ensuring precise model pinning, consented writes, and fail-closed operations. It delegates implementation to specialized AI roles (Astra, Terra) while the human architect retains control over requirements, architecture, and verification, with a read-only advisor (Sol) for final review.
MCP server and CLI tool that gives coding agents a structural map of codebases across 20+ languages using tree-sitter, reducing token usage by up to 75% compared to grep-based exploration.
DSH Mini TUI is a minimalist terminal UI plugin for DeepSeek Harness (DSH). It retains full agent and tool capabilities, but by default keeps execution in the background, providing a lower-noise terminal interface focused on answers.
Loctree is a fast, multi-language static analysis tool and MCP/LSP server designed to provide structural code intelligence for AI agents and developers, offering dependency graph queries, dead export detection, and circular import analysis.
CLI and Claude Code plugin that runs parallel headless Claude sessions, monitors real Anthropic usage, and stops before exceeding your subscription quota. Provides run, process, overview, and connect commands with safety guards.
RAPID is a lean, spec-driven methodology for AI-assisted development, structuring workflows from requirements to delivery with human approval gates.
GeeksPulse is a free, ad-free developer news aggregator that curates 50 hand-picked RSS feeds across 11 categories into a fast static site with a cyberpunk UI, AI summaries, bookmarks and keyboard shortcuts.
Stageflow is an open-source Node.js runtime for configurable multi-stage agent workflows: author YAML pipelines with typed handoffs, DAG execution, human gates, MCP, CI, and a local operator console, running on the Pi coding agent.
Gateship is a local-first delivery runtime for coding agents. It gives a task an isolated Git worktree, runs a local Claude Code or Codex session, verifies acceptance commands, adds an independent read-only review, and ships a squash-merged pull request. State lives in local SQLite, served by a Bun web UI on 127.0.0.1.
Local-first toolkit for shader generation, review, validation, and export to Shadertoy, WebGL, LÖVE 2D, and video workflows. Supports AI-assisted drafting, license filtering, render preflight checks, and web UI with live preview.
VALP is an open protocol and reference CLI for visible, evidence-backed multi-agent work. It audits agent completion claims by checking receipts, expected evidence, review gates, and final synthesis to prevent false-done states.
A Go CLI for isolating and managing multiple Google Antigravity accounts via dynamic $HOME sandboxing, with real-time quota tracking and Herdr multi-agent integration.
DevDNA transforms your public GitHub activity into a developer case file, an evidence badge for your README, and a shareable report. It uses GitHub OAuth for login-only self-analysis, classifies developers into ten profile types, and produces deterministic rule-based summaries.
stepstone keeps a repository's roadmap inside the repository: dependency-aware Project Goals committed alongside the code and driven by any coding agent or human through one CLI, with next, ready and waves answers, an optional terminal board, an Agent Skill, and a Pi extension.
PyFFmpegCore is a safe, explainable FFmpeg task runner for the terminal, Python, and CI, designed to replace fragile command strings with repeatable media automation workflows.
An AI-first engineering toolkit that provides specialized knowledge, workflows, and standards to turn AI coding agents into senior mobile engineers.
A Homebrew tap packaging Agenxy's command-line tools: the Dibs cask (dibs and dibd binaries with a man page), plus Remap and Supgang formulae, including migration entries so older Lanes installs keep updating.
Parker is a minimal, keyboard-first macOS menu-bar editor for notes, code and Markdown, built on Tauri with React and CodeMirror. It works on a plain folder of files you own, with aggressive auto-save and optional built-in git backup for versioning.
An auto-updating index of LLM agent frameworks on GitHub, refreshed every 15 minutes by a GitHub Actions cron job that scans, diffs and republishes a structured table of tracked repositories.
A pre-research hook for Claude Code: it runs Haiku to explore your codebase with Glob, Grep, Read and Git tools, then injects a structured findings block as context before Opus or Sonnet responds, with caching and skip detection.
Web UI Quality is an Apache-2.0 open-source preview CLI that keeps scope, actual change, evidence and verification separate for AI-assisted Web edits, and refuses to treat a model suggestion as write authority.
SupaNexus is an open AI gateway that unifies access to multiple LLM providers via one API, simplifying development with OpenAI-compatible interfaces and centralized model management.
TangleClaw is an AI coding session orchestrator for macOS that wraps Claude Code, Codex, Antigravity, Aider, and OpenClaw sessions in persistent tmux processes. It provides session continuity, structured wrap summaries, searchable history, secure remote access via Caddy, inter-session agent messaging, and a dashboard PWA — all with zero npm dependencies.
jarvis (jarvis-mcp) is a local-first code intelligence MCP server for coding agents: it precomputes SCIP navigation, a Tree-sitter declaration baseline, Zoekt lexical search and optional semantic vectors, then serves nine read-only tools over stdio, keeping all data in a local directory.
Swarm is a local-first control plane for AI-agent development that monitors every Claude Code, Codex, Gemini, Grok, Aider, and opencode session on your machine — tracking live tool calls, token spend, cost, tasks, worktrees, and resources — with rules enforcement, a coordination ledger, and a unified dashboard.
Forgehouse is an open-source, full-stack AI website studio that lets users create, revise, preview, and publish browser-rendered websites through natural-language prompts. It features Groq-powered generation with a safe fallback, immutable revision history, sandboxed live previews, and application-hosted publishing.
PARTHA turns one repository revision into a sealed, queryable intelligence model — building architecture graphs, dependency graphs, engineering reviews, insights, documentation, and optional AI context from a single persistent snapshot with evidence and provenance.
LogSquirl is a fast, open-source log explorer for Windows, macOS, and Linux. Search multi-gigabyte files, follow live logs, and chart patterns. Built with Qt6/C++, it's a fork of klogg and glogg.
A collection of seven skill libraries for AI coding agents covering code cleanup, prose editing, parallel agent orchestration, interface design, product feedback, design research, and QA. Works with Claude Code, Cursor, Codex and other agents that read skill files.
cctrace is a TLS-intercepting tracer for Claude Code, Codex, Grok, and Kimi CLIs that captures every API call in a live web UI with session replay, cost tracking, and cross-project dashboards.
Peaky Peek is a local-first, open-source audit and trust console for AI agents. It captures causal chains, verifies claims deterministically, and produces explainable trust scores with failure narratives — all running on your machine.
A comprehensive kit that transforms a Claude Code subscription into a team-oriented assistant accessible via Telegram, featuring tools for agent monitoring, skill management, and client communication.
lore is a Bun-based command line tool that mirrors Claude Code's local transcripts into an archive, indexes them with SQLite FTS5, and serves a local web explorer for searching sessions and profiling token usage.
Hive Desktop is a local macOS app that collects pull requests, issues, notifications, alerts and incoming webhooks into one queue, then filters and routes them through YAML-defined flows that can also trigger coding agents or shell commands.
MOZAK gives a research topic or code repository a Git-native memory: six modules record goals, evidence, decisions, progress and reusable concepts so people and coding agents share the same project state across sessions.
An AI-powered Git toolkit and MCP server written in Rust, featuring intelligent commit message rewriting, PR generation, and integrations for Atlassian, Datadog, and Google services.
A standalone, type-aware GDScript language server for Godot 4.6 and 4.7, implemented as a faithful Rust port of the GDScript frontend.
UCloud SDK for Node.js: a client library for calling UCloud cloud APIs from Node.js. The README links Chinese-language guides on quickstart, configuration, error handling, middleware and generic calls, plus examples for batch-creating cloud hosts and building a two-tier load-balanced architecture.
CoalLedger is a docs-health checking suite for AI coding agents — the documentation counterpart to CoalMine's code canaries. Seven canaries flag wrong, incomplete, stale, contradictory, broken, unreadable and (optionally) leaked documentation, with a bundled zero-dependency CommonMark+GFM AST engine, context-judged severity and choice-gated fixes.
A fully offline, deterministic MCP server that gives AI agents local tools for exact math, date and timezone handling, unit and base conversion, subnet math, text distance and token counting — no network calls, telemetry or API keys.
A local, read‑only dashboard that visualizes Claude Code usage—token counts, cost estimates, sessions, tools, skills, projects, and activity patterns—by scanning ~/.claude/projects/ and serving a FastAPI‑based UI.
Ghostget is an open-source CLI and SDK that provides AI agents with precise, bounded web capabilities. It enables secure page capture, media archiving, and encrypted reads from connected services like social platforms and email, ensuring local custody and explicit trust boundaries without granting unrestricted browser or API access.
UE5 asset workbench plugin enables scripts and AI agents to read, edit, import, migrate, and audit Unreal Engine 5 uasset files by exporting structured JSON and applying changes via commandlets, supporting Blueprints, animations, widgets, materials, Niagara, levels, etc.
RepoScout is a fast, local-first Rust CLI for repository intelligence, providing code metrics, complexity analysis, and duplication detection for humans and coding agents.
Knot is a management tool for cloud development environments running in a Nomad cluster or locally. It offers a web UI and CLI for deploying, accessing and sharing dev spaces, with templates, quotas, user permissions, browser VS Code, terminals, VNC and KVM virtual machines.
VS Code extension that turns trailing-dash comments into navigable sections, creating a table-of-contents-like outline for large code files across many languages.
A local-first web UI for running Claude Code and Codex sessions in browser tabs with real PTY terminals, persistent history, and agent coordination via MCP. Sessions run in tmux for durability, with token-authenticated localhost security.
A native macOS application for fast video editing and a specialized frame handoff workflow for providing annotated visual bug reports to coding agents.
Cloudx is a local-first mobile and desktop workbench for running and supervising Codex CLI sessions on your own Linux machine: pane layouts, file tools, diffs, worktrees, plugin tabs, a local documentation archive, Jira views and constrained voice control.
Reel generates always-current demo films for web apps and CLIs from a YAML spec. It drives a real browser or terminal, records the actual flow, and renders GIF, MP4, WebM, storyboards, and interactive HTML compositions. Works via CLI, Studio UI, or Claude Code plugin. Fails CI when demos break.
VibeCoder is an automated GitHub issue worker that monitors repositories, picks up issues, writes code using AI agents (Claude, Codex, Gemini), runs quality checks, and opens pull requests. It supports review feedback loops, self-healing, and cost optimization.
Adeorq is a Windows/Linux desktop panel that runs AI coding agents in real terminals, using your existing subscriptions and CLIs. It features a multi-terminal cockpit, session management, canvas, and more, with a source-visible proprietary license.
An unofficial Codex Skill for repository delivery: it routes work between the current task, internal agents and optional App tasks, with explicit ownership, isolated Git worktrees, independent review, an installer with check/dry-run modes, and deterministic tests.
gha-doctor is a zero-config CLI that diagnoses GitHub Actions workflows for flaky jobs, wasted minutes, slow steps, cache problems and anti-patterns, with static lint rules, run-history analysis, a health score badge, auto-fixes, SARIF/HTML/JSON reports, a GitHub Action and an MCP server.
