DEDUPEDHarness2
Harness² is a Python framework from Google Research that refines the harness around a fixed model (prompts, tools, memory, skills, orchestration) so it performs better on agent tasks.
3 use cases: Try both recursion modes locally with no credentials; Refine an agent harness on a small LAB legal set; Compare recursion budgets on WorkBuddy-Bench code tasks · GitHub
iFixAi
iFixAi is an open-source Python tool that audits whether an AI agent is doing the job it is supposed to do, using 60 inspections that combine red-teaming with operational checks.
5 use cases: Audit your agent in a guided terminal wizard; Have Claude Code run and explain the audit; Run a scripted audit in CI with explicit flags · GitHub
Orca
Orca is a desktop app, with a mobile companion, for running several CLI coding agents side by side, each in its own git worktree, and tracking them in one place.
5 use cases: Compare agent results across parallel worktrees; Review AI diffs line by line and send feedback back; Check on and steer agents from your phone · GitHub
Diagram Design
Diagram Design is an agent skill that makes one self-contained HTML and SVG diagram file in a consistent editorial style, matched to your brand.
4 use cases: Draw an architecture diagram for a doc or post; Match diagrams to your site's brand; Redraw old Mermaid or draw.io diagrams · GitHub
AI Memory
ai-memory is a Rust tool that gives AI coding agents a shared long-term memory, stored as a git-backed markdown wiki, so work can be handed off between agent CLIs and machines.
4 use cases: Hand off a half-finished task from Claude Code to Codex; Resume a project on another machine with the same memory; Share project knowledge across a team on one server · GitHub
Graphify
Graphify is a tool that turns a codebase, its docs, configs, and PDFs into a knowledge graph that an AI coding assistant can query instead of grepping files.
4 use cases: Map an unfamiliar codebase in one run; Answer architecture questions from the graph; Trace how two components connect before a change · GitHub
Git AI
gitai is an open-source Git server and CLI, written in Go, built so many AI coding agents can read, write and land code on the same repositories at the same time.
4 use cases: Run a local gitai server for concurrent agent lands; Land conflicting agent changes with a deterministic resolver; Move repository state into SQLite or PostgreSQL and object storage · GitHub
Magpie
magpie is a local gateway and menu bar app that lets agents like Claude Code, Codex and Gemini CLI use models from many providers and subscriptions, with routing and failover.
4 use cases: Run Claude Code on DeepSeek instead of Claude; Keep an agent working when one quota runs out; See what is left on every AI plan in one command · GitHub
Awesome DeepSeek Harness
Awesome DeepSeek Harness is a curated list of plugins, tools, and infrastructure for the DeepSeek Harness (DSH) ecosystem, with install commands for the official runtime and external profile bundles.
5 use cases: Start the official DSH web runtime in one command; Add a community bundle to a DSH web profile; Move Claude Code and Codex sessions into DSH · GitHub
Resimind
ResiMind is an open-source Python architecture that checks agent steps with independent verifiers and reuses checked rules.
5 use cases: Give an existing agent exact algebra checks via MCP; Teach an agent a bounded verify-and-repair workflow; Test whether a math tool rejects a convincing wrong answer · GitHub
Fauxnix
fauxnix is a TypeScript tool that translates Linux-style bash commands into PowerShell so AI agents can run them natively on Windows, without a VM or WSL.
4 use cases: Give Claude Code bash-style commands on Windows; Run Codex bash commands on Windows through Fauxnix; Run a multi-step bash plan in one MCP call · GitHub
Building with Typesafe Jev
A community skill that teaches coding agents to design with TypeSafe AI's Jev model, using typed questions, calibrated confidence, and a library of 150+ linked example projects.
4 use cases: Design a typed Jev ticket router in Claude Code; Build a Jev pull request grader in Codex; Check Jev designs against the live API · GitHub
Pi Agent Desktop
Pi Agent is a local desktop app for macOS, Windows and Linux that packages the pi coding agent into an installable application with a visual interface.
5 use cases: Resume past Pi coding sessions from a desktop window; Schedule a saved prompt to run as a new session; Branch or fork a Pi conversation from any message · GitHub
Acryl
ACRYL is an open source agentic development environment where one persistent workspace holds the project context while any coding agent can plug in and be swapped out.
5 use cases: Keep one project context across coding agents; Run several agents and act only when one needs you; Review agent diffs line by line and send comments back · GitHub
Factory
Factory is a shell-based reference that installs a repeatable, GitHub-backed software delivery loop into an existing project for Claude Code, with a thin Codex adapter.
4 use cases: Install the Factory into an existing GitHub project; Turn small labeled issues into reviewed draft pull requests; Sort an issue backlog into implement, spec, or wait · GitHub
Operator Memory
Operator Memory is a context engine that gives coding agents a Markdown workspace for specs, decisions, standards, and lessons, so each new session starts with what earlier sessions learned.
4 use cases: Give an existing repo a memory the agent keeps; Publish agent-written specs to your team via Git; Carry your personal agent rules into every project · GitHub
ECC
ECC is an MIT-licensed agent harness of skills, agents, hooks, memory, and AgentShield security scanning that runs with Claude Code, Codex, and other coding agents.
4 use cases: Install ECC as a Claude Code plugin in one guided run; Preview, then install ECC into Claude Code and Codex together; Scan agent prompts, hooks, and permissions for security issues · GitHub
Code AF
CodeAF is an open-source Go coding harness for open models that runs as one local binary, with specialist subharnesses for coding, security audits and pull request reviews.
5 use cases: Audit local code changes for security problems; Review a GitHub pull request with findings and reasons; Hand a coding fix to senior-dev on its own branch · GitHub
Agent Launcher
Agent Launcher is a local desktop app that detects, links, configures, and runs existing coding-agent CLIs such as Claude Code, Codex CLI, and Gemini CLI from one workspace.
4 use cases: Run all installed coding agents from one desktop app; Set up an OpenAI-compatible provider profile for Codex CLI; Review local token usage and sessions across agents · GitHub
Agent Chrome Relay
Chrome Relay is a macOS tool that lets coding agents drive your everyday Chrome, signed in as you, in background tabs from the command line.
4 use cases: Fill and submit a web form in your signed-in Chrome; Read a logged-in dashboard page step by step; Sign in to a site with 1Password, keeping secrets from the agent · GitHub
DSH Worktable
dsh-worktable is a Cordis plugin for DeepSeek Harness that adds a sidebar app drawer, a dockable split workspace for each project, and a control room that mirrors the status of every bound project.
4 use cases: Watch every agent project status from one control room; Work on a project in one split window with files, terminal, and browser; Point at a window element and send a fix request to the agent · GitHub
Crux
Crux is a coding agent built on the pi agent, re-engineered for long-horizon terminal tasks, with a Terminal-Bench 2.1 benchmark harness.
3 use cases: Benchmark a self-hosted model on Terminal-Bench 2.1; Test whether an agent change helps, task by task; Find why long agent runs stall, truncate, or time out · GitHub
Res
RES is an agent skill that studies a web SaaS product through the browser and writes a knowledge base of its screens, rules, data model, design tokens and production barriers for rebuilding it.
3 use cases: Turn a SaaS product into a build kit for your team; Find what blocks a clone from launching; Check a knowledge base and build its report and zip · GitHub
Flare
Flare is a graph-first IDE for agentic coding that maps a codebase as a live graph, runs a terminal for agents like claude, codex or opencode, keeps a shadow history of changes, and holds a shared task board and chat for agents.
4 use cases: Hand an agent a task card with the graph context; Review and revert agent edits from shadow history; Keep two agents from colliding on the same files · GitHub
Free Claude Code
FCC is an open-source Python proxy that lets coding agents such as Claude Code and Codex use free, paid, and local models from many providers through one Admin UI.
4 use cases: Run Claude Code on free models through FCC; Keep a coding session running during provider outages; Run a Codex session on a project folder in the browser · GitHub
Awesome Jev
A curated list of open-source projects, demos, SDKs and skills built around Jev, TypeSafe's model for yes/no judgments, choices, and scores.
4 use cases: Lint Jev questions before sending them to production; Check changed code lines against plain-English rules; Flag anomalous fields in data files with a Jev-based CLI · GitHub
Old Coder
A markdown skill for coding agents that makes them write a test plan before coding and produce an evidence report after, so a human reviews those documents instead of the code.
4 use cases: Rerun the demo gauntlet to check the evidence yourself; Build a feature in Claude Code with a spec you approve first; Add the Old Coder workflow to Codex or Cursor rules · GitHub
Lcu
LCU is a tool that lets agents in any harness use Codex's computer-use runtime to read windows, click, type, and take screenshots in desktop apps and Chrome, without Codex authentication.
4 use cases: Let Claude Code drive desktop apps with native approvals; Give Codex CLI control of desktop apps on macOS or Linux; Let an agent read and control Chrome tabs with site approval · GitHub
Intent Router
Intent-Router is an installable skill that turns a vague request for an AI agent into a typed IntentSpec contract, and later checks the delivered work against that contract.
3 use cases: Turn a vague coding request into a checked spec; Delegate a task to an agent and check the result; Stop an agent from guessing past an irreversible step · GitHub
Opencode
OpenCode is an open source AI coding agent that runs as a command-line tool and as a beta desktop app.
5 use cases: Explore an unfamiliar codebase without risking edits; Hand multistep code searches to a subagent; Build a code change with the default full-access agent · GitHub
Awesome Decision Models
A curated, community-maintained list of decision models (models that return option probabilities instead of generated text), with the hosted APIs, open-weight models, runtimes, SDKs, benchmarks, and papers around them.
4 use cases: Route support tickets with a hosted decision API; Run a small open decision model locally; Check agent tool calls before they run · GitHub
Ant Omni Evo
AntOmniEvo is a Python framework that evolves a system's tunable files (skills, prompts, workflow configs, or single-file algorithms) through a loop where AI coding agents propose edits scored by your own evaluator.
4 use cases: Evolve an agent skill directory against your eval set; Optimize a retrieval workflow's pipeline.json and node code; Optimize a single-file algorithm and its description · GitHub
Claude Code
Claude Code is an agentic coding tool that runs in your terminal, reads your codebase, and carries out coding and git tasks from natural language commands.
3 use cases: Explain complex code in an unfamiliar project; Delegate routine coding tasks from the terminal; Handle git workflows with natural language · GitHub
Claude Plugins Official
A curated, Anthropic-managed marketplace of Claude Code plugins, split into internal plugins and third-party plugins.
4 use cases: Install a vetted plugin from the official marketplace; Build a plugin in the standard layout for submission; Publish existing SKILL.md files as a marketplace bundle · GitHub
Claude Quickstarts
A collection of starter projects that show developers how to build deployable applications on the Claude API.
4 use cases: Run a Claude customer support agent over a knowledge base; Analyze financial data by chat with interactive charts; Post a weekday brief of Slack and GitHub activity · GitHub
Pi Book
A Chinese architecture book, with English and Spanish translations, that explains how the @earendil-works/pi-agent-core agent loop library is built, citing the pi source code.
3 use cases: Have an agent verify the book's code citations; Run the book's web reader on your own machine; Compare the Chinese, English and Spanish editions · GitHub
Aoci Code
AOCI-CODE is a local-first MCP server and Go CLI that has your coding agent write a Git-versioned index of your codebase and optional database schema, which agents read before they change anything.
4 use cases: Take over an existing codebase with an agent in one pass; Resume a project in a new agent or chat without starting over; Index PostgreSQL or MySQL table structures next to the code · GitHub
Frontier Agent
FrontierAgent is an open-source agent runtime and terminal app (the frontier-agent TUI) with ReAct and Agent Team workflows for long-horizon research and file-based work.
4 use cases: Research a question with a parallel agent team; Analyze a project or documents with one stateful agent; Install frontier-agent once and launch it in any folder · GitHub
Appllama Skills
A set of agent skills that teach AI coding agents to research top-grossing mobile apps through the Appllama MCP and build native-feeling Expo / React Native screens checked in a simulator.
5 use cases: Build a mobile app's screens from top-grossing patterns; Improve an existing app screen against category leaders; Research how top apps structure onboarding and paywalls · GitHub
Fast Browser Use
Fast Browser Use is a local-first browser automation engine and agent skill that uses open Qwen3.5 models to pick page actions from visible elements.
3 use cases: Check a web settings form saves correctly, offline; Give a coding agent a fast local browser for page tasks; Measure task time for two local Qwen models on your sites · GitHub
Kaggle Tpu Lab
A set of notebooks and scripts that run open AI models on Kaggle's free TPU v5e-8 and expose them as a public endpoint that speaks the OpenAI and Anthropic APIs.
5 use cases: Run a free open model as an endpoint for Claude Code; Point Codex CLI at a Kaggle-hosted open model; Launch a model server from the terminal · GitHub
Rutis
rutis is a plugin runtime, with a Rust core, that lets software and AI agents write, swap and hot-reload plugins in TypeScript, Python, Go and Swift across processes and machines.
4 use cases: Iterate on a plugin without restarting the host; Catch missing service dependencies before a plugin runs; Prototype a capability in Python, then move it to another machine · GitHub
Design How
A set of agent skills and three shadcn-compatible design systems (Glaze, Lime, Quartz) that help coding agents turn an app's existing UI into a design system, migrate screens onto it, and check for drift.
4 use cases: Build a design system from your existing app UI; Move every app screen onto a design system safely; Block new UI drift in CI with a design system check · GitHub
Pairfob
Pairfob lets you control the coding agents running on your computer from a phone, tablet, or another computer, using the same live sessions over an end-to-end encrypted connection with no open ports.
5 use cases: Pair your phone to agents on your computer; Answer agent prompts from your phone; Review agent code changes on your phone · GitHub
Aa Agentperf Local
agentperf-local is an open-source tool that replays recorded agent conversations against a local OpenAI-compatible model server to measure throughput and latency.
5 use cases: Benchmark a local model server you already run; Download and benchmark a pinned catalog model; Check a setup quickly with the mini replay · GitHub
See Code
SeeCode is a JavaScript tool that turns a short JSON spec into an animated, explorable diagram that an AI coding agent can create and export.
5 use cases: Turn a system description into an interactive architecture diagram; Redraw existing Mermaid or draw.io diagrams in See Code style; Export a diagram as SVG, PNG, GIF or MP4 for slides, docs or posts · GitHub
Wikiskill
WikiSkill is a Python framework that lets an AI agent turn its past runs into a persistent wiki and evolve its own skills, keeping only changes that score better on held-out tasks.
5 use cases: Evolve an agent's skills from its own task traces; Test whether an evolved skill actually improved results; Run the WikiSkill loop with Claude Code as the agent · GitHub
Agent Standup
Agent Standup is a local app that reads the session files written by Claude Code, Codex and Mistral Vibe and shows them as a live animated team call and a board of sessions.
5 use cases: Watch your agent sessions as a live team call; Find stuck or looping agents before they waste time; Find which sub-agent uses the most tokens · GitHub
Atria Dawn
Atria Dawn is a 744B-parameter MoE agentic model from the Shanghai Artificial Intelligence Laboratory, with hosted API access and config recipes for Codex, Claude Code, and Kimi Code.
4 use cases: Run Codex CLI on Atria Dawn as a text-only agent; Block Claude Code from reading PDFs and images; Connect Kimi Code to Atria Dawn over Chat Completions · GitHub
Openharness
Harness is a command center that runs and watches coding agents and creative tool agents across your own machines, with a viewer for each craft.
5 use cases: Run coding agents on several machines in one window; Build a circuit and compare two versions of it; Pin a robot moment and test a disturbance · GitHub
Claude DB
claude-db is a free, open-source memory layer for Claude Code that stores decisions, dead ends, rules and a map of your code in your own database.
4 use cases: Pick up a Claude Code project where you stopped; Map a codebase so Claude finds callers in one call; Leave a short handoff note for the next chat · GitHub
Jevmem
jevmem is an open source TypeScript tool that saves decisions, rules and failed approaches from Claude Code, Cursor and Codex chats to a JEVMEM.md file in your repo and brings the relevant lines back in later sessions.
4 use cases: Stop Claude Code from repeating past decisions and dead ends; Move an existing CLAUDE.md into JEVMEM.md lines; Share project decisions with your team through git · GitHub
Axiomcodegraph
AxiomCode Graph is a tool that builds a call-graph knowledge map of a codebase so AI agents and developers can query exact call paths and impact instead of grepping.
4 use cases: Trace the exact call path between two functions; Find what breaks before changing a function; Let an agent explore a new codebase without grep · GitHub
Procoder
Procoder is a Go binary and agent plugin that adds a commit gate, quality controllers, and a lessons loop to AI coding agents.
4 use cases: Gate agent commits so bad changes never land; Stop an agent from closing unfinished tasks; Run release checks before a version tag · GitHub
Quilt
Quilt is a local, single-user AI design tool that generates multi-screen HTML app designs on an infinite canvas and lets a coding agent read and write that canvas over MCP.
3 use cases: Turn an app idea into a clickable multi-screen prototype; Let your coding agent read and edit design screens; Set one brand palette and reuse shared navigation across screens · GitHub
BGTS Context Engine
A deterministic code-graph context engine that indexes your repositories and returns ranked code symbols to AI coding agents over MCP, REST or the command line.
4 use cases: Find the symbols behind a bug question before editing; Let a coding agent pull ranked context over MCP; Replay a retrieval call to audit a bad agent change · GitHub
Optimaizr
A local CLI that finds wasted tokens in Claude Code, Codex, opencode and your own API calls, and fixes some of it.
4 use cases: Find where Claude Code tokens are wasted; Hold expensive Claude Code messages before they go; Track token use in your own Anthropic or OpenAI app · GitHub
Dioramas
Dioramas is a free, open-source three.js framework and set of 20 example sites for building cinematic, interactive 3D landing pages with AI-generated hero models.
3 use cases: Build a cinematic 3D landing page from a brief; Generate hero 3D models from text prompts; Study 20 interactive 3D site patterns for a design brief · GitHub
Pi Herdsman
Pi Herdsman is an orchestration layer that connects Pi coding-agent sessions with herdr so that background, nested, and multi-branch Agents can work in parallel under explicit supervision.
4 use cases: Delegate a background code investigation to an Agent; Split a project into parallel branch work under a Manager; Check token usage and cost for a session and its Agents · GitHub
Boardui
BoardUI is a React design system for agentic interfaces, shipped as source files you copy into your project, with a working AI chat app on your own model key as its homepage.
5 use cases: Deploy a working AI chat app with your own model key; Run the BoardUI chat app on your machine; Copy BoardUI components into an existing Next.js app · GitHub
Botbus Protocol
BotBus Protocol is the public wire contract, Swift connector code and test fixtures behind the BotBus apps, which let you follow and approve Codex and Claude Code tasks from your phone and watch.
4 use cases: Approve Codex and Claude Code requests from your phone; Review agent screenshots, previews and changes on your phone; Pair a Linux machine running agents with your phone · GitHub
Brig
Brig is a Go command line tool and session daemon that runs a coding agent inside a microVM sandbox on your own machine.
3 use cases: Run Claude Code on one project inside a sandbox; Try unfamiliar code in a disposable sandbox; Check what a sandbox forwards before an agent runs · GitHub
macOS Harness
A thin set of shell commands that lets an LLM agent read and control apps running on a Mac, one command per step, in the background.
3 use cases: Let a coding agent operate a running Mac app; Fill fields in a desktop app that has no API; Check what an app window shows after each step · GitHub
SVG Diagram
A house style and zero-dependency linter that lets a coding agent hand-place SVG diagrams and check them against fixed layout, font, colour and arrow rules.
3 use cases: Draw a flowchart or architecture diagram as lint-clean SVG; Lint an existing SVG and read findings as JSON; Draw a sequence or data-flow diagram with CJK labels · GitHub
CTX Handoff Mod
ctx-handoff is a Claude Code plugin that hands long sessions off to a fresh conversation, keeps the cache warm while you are away, and saves what you teach Claude as project notes.
4 use cases: Hand off a full Claude Code session to a fresh chat; Keep a long chat's cache warm while you step away; Turn repeated instructions into project notes · GitHub
Codex Mobile Dev Plugin
Mobile Dev is a Codex plugin that shows iOS and Android devices beside a Codex desktop chat, with device control, combined logs, and performance monitoring for mobile app work.
4 use cases: Run an Expo app in a simulator beside your chat; Debug an app error from combined native and JS logs; Point Codex at a UI element to fix it · GitHub
AI Engineer Notebooks
A set of free, framework-free Jupyter notebooks in Google Colab that teach the applied-LLM stack for AI Engineer and Forward Deployed Engineer roles, running on the free Groq API.
5 use cases: Build a RAG question-answering prototype with a golden set; Catch quality regressions before a prompt or model change ships; Test an agent against prompt injection before it touches real data · GitHub
Slotstream
Slotstream is a native Swift command-line tool and local server that runs the 125-billion-parameter Qwen3.8-Flash-Next model on Apple Silicon Macs with 16 to 64 GB of memory by streaming experts from the SSD.
4 use cases: Chat with a large model offline on your Mac; Code with Claude Code on a local model; Use Codex with a local model on your Mac · GitHub
Cjkfold
A Python library and command line tool that folds look-alike CJK radical glyphs back to the ordinary Han characters they imitate, and reports look-alike and invisible characters in text.
4 use cases: Check PDF-extracted Chinese text for look-alike glyphs; Fold a RAG corpus and its queries so Chinese terms match; Audit a folder of extracted text for hidden characters · GitHub
Context Attachment Manager
A Codex and Claude Desktop plugin that lets you choose which past images are sent to the model with your next message, so long conversations stay small.
4 use cases: Keep a long Codex image task going by unchecking old images; Cut Claude Code session size in Claude Desktop by leaving out past images; Let the model fetch only the images it needs with Auto-select · GitHub
DeepSeek Harness Desktop
DSH Desktop is an open-source desktop host that runs the official DeepSeek Harness Web UI and runtime in a desktop window, with native iPhone and Android companion apps for continuing work from a phone.
4 use cases: Run DeepSeek Harness in a desktop window; Continue Harness sessions from an iPhone on Wi-Fi; Track Harness token use and estimated cost on macOS · GitHub
Tardigrade
Tardigrade is a TypeScript framework for building composable actors and agents around an immutable event log, built on Effect TS.
5 use cases: Scaffold a research agent from the quickstart template; Rebuild agent state by replaying an event log; Deploy an actor to Cloudflare Workers · GitHub
Awesome Jev
A curated list of open-source projects, SDKs, and integrations built around Jev, TypeSafe AI's model for typed, probabilistic decisions.
5 use cases: Rank a shortlist by weighted criteria in CI; Label DataFrame rows with closed-set Jev answers; Find repository files by meaning, not keywords · GitHub
PS5 LM
PS5LM is an open-source llama.cpp app that runs GGUF language and decision models on a jailbroken PS5's GPU and serves them over the local network.
4 use cases: Run a private model server on your PS5; Point Claude Code at your PS5 model; Classify support messages with a decision model · GitHub
Grasp
Grasp is a self-hosted workspace that runs coding agents inside Docker sandboxes through visual FSM workflows with human approval gates.
5 use cases: Run a coding task from one sentence to a reviewed change; Design a custom workflow with fail and rollback paths; Review agent specs and diffs with an outside reviewer · GitHub
Abide
Abide is a CLI that checks a coding agent's edits and turns against the project's own instruction rules and makes the agent repair any violation.
4 use cases: Stop a coding agent from breaking your written project rules; Find every existing rule violation in a codebase at once; Switch off or rewrite rules that never fire or fire wrongly · GitHub
Orca Replay
OrcaReplay is a TypeScript npm tool that records an AI agent run through a local proxy and lets you replay it offline, fork it onto other models, and inspect what happened.
5 use cases: Find the exact step that broke an agent run; Fork a run at one step and compare models; Try Orca Replay with a sample run and no key · GitHub
Agent Foundation
Agent Foundation (a13n) is an open-source, self-hosted library and platform for building and running your own AI agent systems, with managed agents, memory, sandboxes, computer use, and durable execution.
4 use cases: Run your own agent service and try an agent; Embed reusable agents in your own application; Try an agent on your own project in a terminal · GitHub
Diri
diri is a native macOS app, with Linux in beta, for running many terminal coding agents in parallel, seeing which ones need you, and reviewing their changes in one place.
5 use cases: Track which of many coding agents needs your input; Give each coding task its own Git worktree and branch; Have one agent spawn and coordinate other agents · GitHub
Star CLI
A TypeScript command-line interface for AI agents that connects to several LLM providers, runs file, shell, and web tools behind a permission gate, and saves sessions.
4 use cases: Review working-tree changes and draft a commit message; Plan a change read-only before any file is edited; Call the agent from a shell script with print mode and JSON · GitHub
Axuse
axuse is a zero-dependency MCP server that lets AI agents drive a Chromium browser (and, on Windows, the desktop) through the accessibility tree and CDP instead of screenshot loops.
4 use cases: Find a fact on a web page and confirm its source; Fill a long web form in one call and submit it; Pull matching lines from a long or endless web page · GitHub
Plugins
A GitHub repository of Cursor plugin specifications and official plugins, where each plugin is a standalone directory with its own manifest.
4 use cases: Scaffold and validate a new Cursor plugin; Split a large task across parallel cloud agents; Review a pull request diff grouped by importance · GitHub
Sure Forge
SureForge is a plain-text, instruction-only Agent Skill that tells an AI agent to research, ask, plan, verify, and get independent review before it calls complex work done.
4 use cases: Plan a multi-step change before any code is written; Run a high-stakes change in full mode with extra checks; Check that an installed Sure Forge copy matches the source · GitHub
Squadra
Squadra is a free, local, MIT-licensed team toolkit that coordinates Claude Code, Codex and Cursor windows with shared tasks, proofs, reviews and a local dashboard.
4 use cases: Coordinate Claude, Codex and Cursor windows on one project; Pick the next window for a task by token use and quota; Prove each change with parallel tests and a source-bound proof · GitHub
Rsi Eval
A lightweight Python SDK and CLI for evaluating recursive self-improvement methods across agent benchmarks.
4 use cases: Preview benchmark tasks before spending model quota; Run a Codex candidate on one Terminal-Bench task; Build your own self-improvement loop around evaluate() · GitHub
Cloudroom Core
Cloudroom core is a self-hostable Rust service that runs coding agents such as Claude Code, Codex, and Pi on a Linux machine and saves their chat history.
4 use cases: Run coding agents on your own Linux server; Build and test the Cloudroom core from source; Back up saved agent chat history · GitHub
Cloudroom Gui
Cloudroom is a desktop app for running Claude Code, Codex, and Pi agents on your Mac or in the cloud, using your own subscriptions.
4 use cases: Move a running agent task to the cloud before you leave; Try several fixes at once in isolated cloud branches; Let a cloud agent read files on your Mac with set limits · GitHub
Fde Gym
FDE Gym is an open online judge where each case is a simulated customer delivery that you work with a coding agent and are scored on once your system is live.
4 use cases: Practise customer delivery cases on your own machine; Run a private FDE Gym site for your team's engineers; Work an FDE Gym case from the command line with the harness · GitHub
Solid Glass
Solid Glass is a TypeScript design system of CSS and React components for AI applications, with a glass-inspired material that works alongside shadcn/ui.
4 use cases: Apply the Solid Glass look to a shadcn/ui project; Add agent status and chat components to an AI app; Run the Solid Glass site to preview components and blocks · GitHub
DeepSeek Harness
DeepSeek Harness (dsh) is an open-source agent harness built on an everything-is-a-plugin architecture using Cordis, with a local Web UI.
4 use cases: Start the DeepSeek Harness Web UI with one command; Build and run the harness from a git checkout; Have a coding agent make a change under the repo's rules · GitHub
Deer Workflow
deer-workflow is an open-source TypeScript runtime for building observable, reusable Agent graphs, where code defines the steps and Coding Agents do the work inside each step.
5 use cases: Generate a runnable TypeScript workflow from plain English; Run a workflow in CI and parse its JSON event stream; Run the Deep Research example to produce an HTML report · GitHub
Open Work Proof
OpenWorkProof is an open protocol that records who authorized an AI agent's work, what the agent did, how it was verified, and who accepted or rejected it, packaged as signed evidence that third parties can check offline.
5 use cases: Verify an agent work package offline without the vendor; Run the Human Agency example to see reserved actions; Attach a four-question evidence report to pull requests · GitHub
Jev Review
A small TypeScript code-review workflow built on TypeSafe Jev that reviews a Git diff or scans a codebase and shows the findings in a local dashboard.
4 use cases: Review uncommitted Git changes before committing; Scan a whole package for review risks; Feed a diff review as JSON to a coding agent · GitHub
Claude Unlimited
Claude Unlimited is a local daemon that pools Claude, ChatGPT/Codex and Anthropic API accounts so a Claude Code session moves to the next account when one hits its usage limit.
4 use cases: Keep a Claude Code session going past a usage limit; Spread a Claude Code session's subagents across accounts; Track AI spend by model and project in one dashboard · GitHub
Agentbox
Agentbox is a self-hosted browser workspace that runs Claude Code and Codex CLI in Docker on a Linux server you control, with chat, terminal, files, Git, and usage tracking.
4 use cases: Run Claude Code in a browser workspace on your server; Review agent changes per file and open a PR; Share agent accounts with a team and track usage · GitHub
Skillranker
SkillRanker is a Rust CLI (`sr`) that uses a Jev-compatible provider to rank agent skills for the next step, based on live session context.
5 use cases: Suggest the right agent skill for each Claude Code step; Inspect a labeled skill ranking offline first; Find out why an expected skill was excluded · GitHub
Omni Route
OmniRoute is a free, MIT-licensed AI gateway that gives you one endpoint for many AI providers and models, with quota-aware fallback between them.
3 use cases: Keep coding when one provider's quota runs out; Send one API call to many models from a script; Compare free-tier capacity before choosing providers · GitHub
Uni Switch
uni-switch is an open-source desktop app for Windows, Linux and macOS that manages API provider settings for Codex and Claude Code clients.
4 use cases: Switch Codex and Claude Code to a new API provider; Sync a provider's models and set context limits; Edit ZCode, DSH or WorkBuddy config files with backups · GitHub
LLM Visuals
A Rust terminal dashboard that reads a locally running LLM server and GPU drivers and shows live throughput, latency, GPU load, context fill and expert routing.
4 use cases: Watch tokens per second and TTFT on a local server; Preview every dashboard view with no GPU or model; Record every request to a SQLite log for analysis · GitHub
Durable Pi
Durable Pi is a set of tools for running persistent AI agents on your own machines, with a web hub, a terminal client and a curl-friendly HTTP API.
4 use cases: Run a durable agent on your own machine; Watch a proactive agent open an incident; Attach a terminal to a running agent and steer it · GitHub
Session Kit
Session Kit is a local terminal picker that lists, names, numbers, and reopens Claude Code and Codex sessions that keep running on the host.
4 use cases: Find the one AI coding session that needs you; Keep AI coding sessions running through SSH drops; Run each Claude Code session on its own account · GitHub
Awesome DeepSeek Harness
A curated GitHub list of plugins, skills, MCP servers, profiles, orchestrators and UIs for DeepSeek Harness (DSH), DeepSeek's agent runtime.
5 use cases: Review and approve agent edits before keeping them; Keep notes and facts across DSH coding sessions; See what fills the DSH context window · GitHub
Bladebro
Bladebro is a Rust browser driver that gives an AI agent full control of a real browser through five MCP tools.
5 use cases: Collect a whole infinite-scroll feed in one call; Run a multi-step web workflow in one agent call; Let an agent use your own logged-in browser · GitHub
Alphacode
AlphaCode is a free, MIT-licensed AI coding agent written in Rust that runs in the terminal and can edit code, run commands and tests, drive Firefox, and control desktop apps.
5 use cases: Fix a failing test or bug in a repo from the terminal; Test a web app flow in a live Firefox session; Drive a native desktop app with a written instruction · GitHub
Enterprise System Design
A source-grounded course and technical reference on enterprise system design, distributed systems, AI systems, and reliable Azure operations.
3 use cases: Design a reliable service with failure and capacity notes; Trace permission-aware retrieval in an AI feature; Test an AI agent's tool authority with security principles · GitHub
Wu
Wu is a fast, native code editor written in Rust, with coding agents built in.
3 use cases: Edit a large codebase in a fast native editor; Hand a coding task to a built-in coding agent; Check Wu's memory use against your current editor · GitHub
Kander
Kander is a kanban board, run from the terminal, that coordinates multiple AI coding agents in isolated git worktrees and gates their work through independent reviewer agents.
4 use cases: Run several coding agents on one backlog in parallel; Block agent changes that fail an independent review; Import GitHub issues as board cards and track them · GitHub
Jevgrep
Jevgrep is a TypeScript CLI (`jg`) that answers repository questions by returning relevant files, reading leads, and verbatim source excerpts for coding agents, using Jev to judge relevance.
3 use cases: Find the code behind a behavior before an agent edits it; Locate the tests that cover a specific behavior; Check which files a search will send before running it · GitHub
Waku
Waku is a fast, native desktop app written in Rust that gives you one place to run and manage local coding agent CLIs, projects, and sessions.
5 use cases: Run several coding agents side by side in one app; Steer an agent mid-task with queued follow-ups; Rewind a Git-backed agent task to a checkpoint · GitHub
Antirion
Antirion is an open-source AI gateway written in Rust, shipped as one binary with a built-in web console, that routes, secures, budgets and observes LLM calls across many model providers.
5 use cases: Send every app's LLM calls through one endpoint; Cap LLM spend per team, project or key; Find out why an LLM call failed or cost too much · GitHub
Harness Mix
Harness Mix is a local kernel that adds a native Codex Desktop interface for connecting many coding agent harnesses and passing tasks between them.
3 use cases: Hand off a coding task between harnesses in one window; Run a multi-harness agent team on a code review; Install one skill or MCP server for many harnesses · GitHub
Enroute
Enroute is an early, open source, programmable Git server written in Rust that stores repositories in object storage such as S3 and exposes hooks and a gRPC API.
4 use cases: Run a local Git server and push a repo to it; Build a code-hosting app on Enroute with a coding agent; Block force pushes to main with a pre-receive hook · GitHub
Agentplay
A local, open source tool that replays coding agent sessions as a map and a timeline, showing what the agent read, edited, and retried.
4 use cases: Check what an agent read before it edited a file; Find the commands an agent kept retrying and failing; Share a session map without prompt text or file contents · GitHub
Simfleet
simfleet is a macOS control plane that runs and tracks parallel React Native work across iOS simulators, Android emulators, Metro ports, and native builds, with a live dashboard and agent device attribution.
4 use cases: Run several app branches side by side without port clashes; Share one native build across agents on JS-only branches; See which coding agent is using which simulator · GitHub
Codex on Crack
Codex on Crack is a JavaScript package that coordinates your configured AI models from inside Codex, handling planning, delegation, host tool handoffs, and review across workspaces.
4 use cases: Plan a project and agree on which model does each part; Have Opus plan and a native worker build, then review; Build with Opus through Claude Code, then check in Codex · GitHub
Skill Corpus
SkillCorpus is an open-source Python pipeline that collects SKILL.md files from public repositories, curates and scores them, and provides retrieval models and evaluation tools for matching skills to agent tasks.
4 use cases: Fetch task-specific skills from SkillHub for an agent; Install the Skill Corpus plugin in your agent host; Build a private skill corpus from your own registry · GitHub
Swe Sweep
SWE-sweep is a Python benchmark harness where an agent must find and fix as many bugs as it can in a real repository, with no hints about bug type or location.
3 use cases: Score one agent patch on a SWE-sweep task; Summarize graded trials by task and in total; Check the harness tests before editing it · GitHub
Tracecrate
TraceCrate is a local-first, client-side workbench that reads AI agent trace files and shows a searchable timeline, recorded metrics, heuristic diagnostics, and side-by-side comparisons.
5 use cases: Read a Claude Code or Codex session step by step; Compare two agent runs event by event; Find failing and slow tool calls in an agent run · GitHub
Termish
Termish is an open-source Android and iOS terminal app for SSH and Mosh sessions, with a Herdr workspace, optional native agent chat, SFTP files and a remote-screen view.
5 use cases: Run Codex in Herdr from your phone; Chat with a coding agent in native phone chat; Keep a remote terminal task alive on a weak network · GitHub
Robot Heads
robot-heads is an MIT-licensed React component library that draws animated 3D robot heads with LED faces showing an AI agent's state, such as thinking, searching, or speaking.
4 use cases: Show a live agent status in a React app; Use small robot avatars in a chat or user list; Review all nine robot states before choosing a design · GitHub
Finn Loop
Finn-loop is a set of three Claude Code skills that use Linear and GitHub to run a small, human-gated loop from spec to build to review, with a human making the merge.
4 use cases: Turn a rough idea into an agent-ready Linear issue; Have Claude build queued issues into pull requests; Get a verdict on every open PR before merging · GitHub
Open Spec
OpenSpec is a spec-driven development framework, installed as a CLI, that gives AI coding assistants a lightweight Markdown spec layer so humans and agents agree on what to build before code is written.
3 use cases: Plan a feature with an AI before any code is written; Switch to the expanded workflow and archive many changes at once; Share one plan across several repos with Stores · GitHub
Pi Posthorse
Posthorse is a TypeScript extension for the Pi coding agent that starts fresh context windows for long tasks and keeps earlier conversation recoverable, without a model-written summary.
3 use cases: Keep a long Pi coding task going across fresh windows; Find an earlier decision in past Pi sessions; Test Posthorse in one Pi session before installing · GitHub
Dalaran
Dalaran is an Apache-2.0 robotics visualization and data tool that logs, stores, replays, and queries multimodal time-series data such as ROS 2 topics, LiDAR, camera, IMU, and TF.
5 use cases: Open an existing .rrd recording in the Dalaran viewer; Save a Python sensor run to a .dlr file for later review; Check a workstation for viewer and driver problems before a robot session · GitHub
Affectui
affectUI is a free, open source library of handcrafted React components, illustrations and motion effects, each delivered as one .tsx file with its styles inside.
4 use cases: Add a polished message action bar to an AI chat; Let an AI coding agent find and add UI items for you; Build a review inbox where humans approve agent actions · GitHub
Flopwire
Flopwire is an open source, self-hostable tool that lets AI agents message each other and search shared history of their past and live sessions across harnesses, machines, and teammates.
4 use cases: Trace a commit back to the agent session that wrote it; Search every past agent session for an earlier decision; Have two agents on one machine agree a work split · GitHub
Threadport
ThreadPort is a local tool that finds unfinished AI coding tasks, reviews their observed evidence, and prepares a continuation for Claude or Codex.
3 use cases: Resume an interrupted Claude or Codex coding task; Hand off a Claude session to Codex through the CLI; Import a Cursor transcript as a paused handoff · GitHub
Eval
A public results set and task definitions for comparing coding-agent harnesses that run the same Kimi K3 model on 30 software-engineering tasks, plus a workflow for evaluating your own harness.
3 use cases: Compare 12 coding-agent harness results side by side; Score your own coding harness on the 30-task set; Check if a harness score is comparable before citing it · GitHub
Zoetrope
zoetrope is a read-only terminal and browser viewer that draws a Claude Code or Codex session as a live flow graph of agents and their tool calls.
5 use cases: Follow a running agent session as a live graph; Replay a finished agent run to review its subagents; Print a session's agent tree without opening the TUI · GitHub
Fuxi
FuXi is a fast terminal AI coding agent that reads code, edits files, runs commands, and drives tools, with cost-aware routing across LLM providers.
4 use cases: Fix failing tests in a repo from the terminal; Run FuXi on your own provider key and switch models; Set up FuXi on a headless machine with a setup token · GitHub
Underclass
underclass is a local Rust proxy that pools multiple ChatGPT/Codex, GitHub Copilot, and Claude Code subscriptions behind one OpenAI-compatible and Anthropic-compatible endpoint.
4 use cases: Point opencode at a pool of coding subscriptions; Route Claude Code through a pool of Claude subscriptions; Watch pool quota, cooldowns, and plans in real time · GitHub
Git Bot
GitBot is a local web app that runs Claude Code, Codex, or OpenCode on your machine as reusable bots with persistent threads and permissions you control.
4 use cases: Flag docs that no longer match the code; Ask Codex about a repo without allowing edits; Share a tested bot with a teammate by code · GitHub
Semantix
Semantix is a Go coding agent and memory kernel that stores reusable slices from past sessions and reduces repeated model calls through caching.
4 use cases: Carry project conventions into every new coding session; Measure how often memory reuse hits on your sessions; Route an OpenAI-compatible client through the gateway · GitHub
Artemis
ARTEMIS is an open source Python tool that turns natural-language instructions into automated tests and tasks on real Android phones and emulators.
4 use cases: Run a phone test from your AI IDE chat; Check a phone setting from the terminal; Assert phone flows inside a pytest suite · GitHub
Graphicle
Graphicle is a local Rust program that builds a code graph of your repository and keeps its docs, decisions and past agent sessions as searchable memory for coding agents.
4 use cases: Give your coding agent the code graph and project memory; Find where a behavior lives by asking from the terminal; See what depends on a function before you change it · GitHub
Ownmem
OwnMem is an open-source, Git-native project memory layer that stores reviewable Markdown in a repository and gives coding agents deterministic, local recall.
4 use cases: Share one project memory across coding agents; Check past failures before changing code; Capture a hard-won lesson as reviewed memory · GitHub
Thinking Orbs
ThinkingOrbs is a Swift package that provides dotted, 3D-looking loading indicators for AI and agent interfaces in SwiftUI.
4 use cases: Show live agent status in a SwiftUI chat app; Show a listening indicator during voice input; Draw orb geometry in SpriteKit, Metal or another surface · GitHub
NIGHTRUN
NightRun is a Rust runtime that boots a PC or Raspberry Pi 5 straight into a local LLM chat from a USB stick or SD card, with no conventional operating system underneath.
3 use cases: Boot an offline LLM chat from a USB stick; Test a NightRun image in QEMU before flashing; Boot a local LLM on a Raspberry Pi 5 from SD · GitHub
DGX Spark Qwen38
A one-command installer that sets up a private AI server on a DGX Spark or ASUS Ascent GX10, serving local text, image, video and speech models through a cockpit and OpenAI- and Anthropic-compatible APIs.
4 use cases: Run a private coding agent on your own DGX Spark; Point Claude Code or Cursor at a local model; Make images and videos with sound on your own box · GitHub
Headroom
Headroom is an open-source tool that compresses tool outputs, logs, files, RAG chunks and chat history on your machine before they reach an LLM, available as a Python or TypeScript library, a local proxy, and an MCP server.
4 use cases: Cut Claude Code token use with one wrap command; Shrink messages in a Python app before each LLM call; Route any OpenAI-compatible client through a local proxy · GitHub
Claude Skins
skins is a Claude Code plugin that redraws tool calls, diffs, tables, code, charts and math in themed cards, and lets you switch or design the look with /skin.
4 use cases: Install skins and pick a readable theme in Claude Code; Have Claude design a custom skin from a short description; Read Claude's Mermaid charts and math as drawn cards · GitHub
Effortless
effortless is a Claude Code plugin that picks the reasoning effort for each prompt and shows the prompt cache countdown, and textless is a desktop app that shows pixel art while Claude works.
4 use cases: Cut Opus coding costs with automatic effort picking; Compact or hand off a heavy chat before the cache goes cold; Show pixel art instead of the step log while Claude works · GitHub
Sloppy
Sloppy is a deterministic, local PHP static analyser that flags code patterns that cost maintenance, with hooks for Claude Code, Gemini CLI and Codex, CI annotations, a Pest plugin and an MCP server.
4 use cases: Catch agent-written PHP debt before the edit lands; Gate CI on new PHP debt only; Review a branch for new PHP debt before merge · GitHub
Builder
AI Builder is an open-source, Apache 2.0 licensed web app where you describe a website or app in plain language and the AI builds, previews, and publishes it.
4 use cases: Build a website from a plain-language description; Start from a template and adapt it to your needs; Let the AI use your other services via MCP · GitHub
Hi Route
HiRoute is a local model router that sends routine agent work to economical models and keeps stronger models for hard tasks, while managing subscriptions, API keys, sessions, usage and cost in one place.
4 use cases: Cut Codex and Claude Code costs with difficulty routing; Route drafting and review to separate model groups; Delegate independent tasks to other agents through a Worker plan · GitHub
My Free Code
My Free Code is an open-source Python gateway that lets Claude Code and other coding agents send requests to many AI model providers and local models through one local address.
4 use cases: Run Claude Code on other model providers; Keep agent sessions running when a provider fails; Run coding agents against local models · GitHub
CLI Anything
CLI-Anything is a Python project that generates command-line interfaces for software so AI agents can operate that software, and it runs a CLI-Hub registry for installing community-built CLIs.
4 use cases: Install agent-ready CLIs from CLI-Hub with pip; Let an agent build and edit FreeCAD models from the terminal; Search and convert a Calibre ebook library with an agent · GitHub
Peerbridge MCP
PeerBridge MCP is a local-first, auditable control room where Codex, Claude Code, Grok, Kimi, provider APIs, and local models work as one governed team on code review and remote work.
4 use cases: Run the provider-free coordination demo; Run Implement and Review on one repo from the Windows app; Continue an earlier agent conversation in a writable room · GitHub
Mobilecode
MobileCode is an open source fork of opencode, an AI coding agent, that detects iOS and Android projects and previews them in a simulator or emulator beside your session.
3 use cases: Preview a React Native app on iOS and Android at once; Let an agent fix a failing mobile build on its own; Run MobileCode from source to change it · GitHub
Humanize
humanize (hmz) is an agent flow system, written in Python, that runs flows of coding agent CLIs such as Claude Code or Codex from a terminal, a browser, or a desktop window.
3 use cases: Run a one-off agent flow from your shell; Watch every epic on your machine in a browser; Try the terminal interface without installing · GitHub
Hycli
Hycli is a Rust tool that turns website interfaces into typed CLI commands and MCP tools for AI agents, with a local dashboard.
3 use cases: Turn a website into a typed CLI in one command; Let an AI agent draft a post with write approval; Check whether a site connection is still signed in · GitHub
KADATH
KADATH is a Python runtime that evolves populations of autonomous AI agents over repeated, graded epochs to optimize a goal measured by a locked benchmark.
3 use cases: Evolve a better agent for a measurable goal; Score agent output against a locked, auditable rubric; Run a web research goal with evolved browsing agents · GitHub
Claude Show Your Work
A Claude Code plugin that shows images, plays videos and runs visual QA with Playwright recordings inside the terminal.
4 use cases: Show a bug fix as a before and after recording; Compare several design directions side by side; Play a scroll animation or ad video in the chat · GitHub
Codex Trajectory
Codex Trajectory is a privacy-aware plugin that turns local Codex task logs into a searchable event ledger, an interactive timeline, and timing, token, and cost views.
4 use cases: Find where a Codex task spent its time; Check token use and estimated cost of a Codex task; Locate failed and slow tool calls in past Codex runs · GitHub
Agentic Playwright
A Playwright and TypeScript test scaffold with page objects, API contracts, data factories, CI, and AI assistant rules that load automatically.
5 use cases: Get a green Playwright TypeScript suite in under 5 minutes; Have an AI assistant write tests that follow team rules; Add API contract tests with Zod validation and fixtures · GitHub
vLLM RLT
vllm-rlt is a standalone inference and serving engine for recurrent language models such as Ouro-1.4B, Nanbeige4.2-3B, and Huginn-0125.
4 use cases: Serve a looped model behind an OpenAI-style endpoint; Compare fixed-depth and early-exit decoding on Ouro; Generate text for a prompt list offline in Python · GitHub
Diff Tour
diff-tour is a read-only terminal app that turns a git change into a guided tour explaining what the change is for and what to check.
3 use cases: Learn why a branch change is written that way; Tour uncommitted changes before you commit them; Reopen a saved tour without waiting for the LLM again · GitHub
Inkloom
Inkloom is a TypeScript codebase for a planned set of AI logo design models, with early access accounts and credits running today while generation is not live yet.
4 use cases: Audit the security and data-handling claims in the source; Model a race-safe code redemption test on the published one; Check what early access does today before signing up · GitHub
Lody iOS
Lody iOS is an independent, community-maintained native iPhone client for Lody that provides chat, session management, and coding-agent collaboration.
5 use cases: Install the Lody iOS beta on iPhone to use Lody on mobile; Review what a coding agent changed, file by file, from iPhone; Open a local dev server running on your Mac from your iPhone · GitHub
Awwwards MCP
awwwards-mcp is a free, open-source MCP server that lets AI coding agents search award-winning websites for design inspiration and analyze, capture, and verify live sites.
4 use cases: Build a landing page from shortlisted award-winning references; Score your build against a reference site and fix the gaps; Turn a site's scroll animations into a runnable GSAP stub · GitHub
System One Connector
A Go MCP connector that gives AI agents direct access to System One models, mainly TypeSafe's Jev, for quick yes/no, choice and score judgments with probabilities.
4 use cases: Triage support tickets by urgency and owning team; Classify a dataset of records in one call; Let an agent act only when it is confident · GitHub
Llam Ampere
llamAmpere is a C++ fork of llama.cpp tuned to run Qwen3.8-27B on Ampere NVIDIA cards such as the RTX 3090 and 3090 Ti, with a compressed KV cache and an MTP drafter.
3 use cases: Run a local 262K-context model server on a 24 GB card; Run 204,800-token context on a 12 GB card; Speed up repeat evals with a prebuilt prompt cache · GitHub
Arena Skill
A free Claude Code skill that runs 100 sub-agents of the same model on one task, pairs their solutions in a judged bracket, and returns the single survivor.
4 use cases: Get a stronger answer after Claude keeps missing; Stress-test a fix for a flaky test; Price a 100-agent run before you spend tokens · GitHub
Tiny Engineer
Tiny Engineer is an open source desktop robot with servos, an OLED face, and a speaker that physically acts out what a coding AI agent is doing, driven over Wi-Fi HTTP or USB serial.
4 use cases: Show your coding agent's progress on a desk robot; Send poses and speech to the robot from any script; Control the robot over USB serial instead of Wi-Fi · GitHub
AI Data Extractor
A free, standard-library-only Python toolkit that extracts your own local chat history from AI coding assistants into normalized JSONL files.
4 use cases: Back up AI coding chat history before it is lost; Turn your coding chats into a fine-tuning dataset; Measure your AI coding usage across tools and projects · GitHub
Percho
Percho is a free, open-source desktop GUI that runs the Pi coding agent's engine in a visual chat interface with tool approvals and themes.
5 use cases: Install Percho and start your first agent chat; Continue a Pi CLI session in the Percho window; Approve or deny each agent tool call visually · GitHub
Agent Hud Open
Agent HUD is an open-source macOS app that shows usage, quotas, tokens and sessions for Claude Code, Codex and many other coding agents at the Mac notch or a screen edge.
4 use cases: Watch every agent's quota at the Mac notch; Answer agent approvals and questions from the HUD; Audit token spend by agent, model and session · GitHub
Evo Trace
EvoTrace is a local-first compiler that turns real Claude Code and Codex sessions into evidence-backed training data, replayable coding tasks, and verified execution rewards, built on DeepSeek Harness.
4 use cases: Find the worth-keeping sessions in your coding history; Compile a session into a Docker-validated coding task; Make a verified task harder and measure its difficulty · GitHub
First Pass
first-pass is a set of rules, skills and checks for coding agents that makes them review the code around a change, prove their work, and fix bugs as a class.
4 use cases: Check your own branch before asking for review; Fix a bug and every same-pattern copy in one pass; Answer ten pre-code questions before writing a feature · GitHub
Pixelcrabs
PixelCrabs is an open-source, local-first visual AI development workspace built on OpenCode, with a Web-first local runtime for previewing and iterating on real project source.
3 use cases: Preview a static site on localhost; Run the pinned OpenCode engine on a project; Run public runtime tests before changing the runtime · GitHub
Lemmalog
Lemmalog is a Rust Datalog engine that gives LLM agents a deductive memory, exposed as an MCP server, a headless CLI, a REPL and an agent skill.
4 use cases: Give Claude Code a persistent, queryable memory; Turn Claude Code into a disciplined investigator; Feed facts into memory from scripts and cron · GitHub
Eval Forge
EvalForge is a Python tool that turns AI evaluation results into portable evidence, applies versioned release policies, and writes JSON, JUnit, SARIF, and Markdown reports for CI.
4 use cases: Block an AI release when eval scores regress; Gate promptfoo eval results before a release; Enforce thresholds on Ragas or DeepEval scores · GitHub
Xyne Spaces
Xyne Spaces is an open source org OS in TypeScript that brings an organization's conversations, documents and tickets into one permission-aware context store, with collaborative apps and sandboxed agents built on top.
4 use cases: Run the full Xyne Spaces stack on your machine; Triage incoming support tickets with Automations; Have a Claw agent review a repository with approval gates · GitHub
Tracely AI
Tracely is a self-hostable tool that grades AI agent traces, clusters failures, freezes bad runs into replayable regression tests, and blocks pull requests that would reintroduce them.
6 use cases: Turn a failing agent trace into a PR-blocking test; Cluster broken agent runs into one issue per failure; Get a Slack or email alert when a live agent run fails · GitHub
SOMA
SOMA is a desktop app for designing parametric 3D models in CadQuery Python code, built together with a CLI coding agent of your choice.
5 use cases: Design a printable case from a plain description; Make a model's dimensions adjustable without code; Check a part for overhangs and thin walls · GitHub
Awesome Claude Code Mods
A community catalogue of 5,000+ Claude Code mods, with the access each mod reports (read, write, run, network) before you install it.
4 use cases: Find a Claude Code mod that fits a task; Keep the prompt cache warm during breaks; Watch context, cost and subagents in a live dashboard · GitHub
Cache Tax
Cache Tax is a Claude Code mod that keeps the prompt cache warm during breaks and shows the estimated rewrite cost before a cold send.
3 use cases: Keep Claude Code's cache warm during a break; See the rewrite price before a cold send; Keep every session warm with always mode · GitHub
Agy Staff
agy-staff is a plugin that lets Claude Code, OpenAI Codex, Pi, and OpenCode delegate review, research, and implementation work to Google's Antigravity CLI (agy).
4 use cases: Get a second-model review of a pull request; Stress-test a plan or design doc before building; Map how a feature works across an unfamiliar repo · GitHub
Bettercode
BetterC0de is a beta desktop app that runs coding agents such as Claude Code, OpenAI Codex, Cursor Agent and Grok in one workspace with an editor, terminal, Git tools and browser previews.
4 use cases: Review agent edits in Git and Diff views before committing; Compare Claude Code, Codex and Cursor on one project; Preview a web UI and send visual feedback to the agent · GitHub
Hermes Jev Skills
A set of Jev-powered decision tools (model routing, memory filtering, skill selection, web screening, mailbox sorting, and more) that plug into Hermes agents and also install for Claude Code and Codex.
4 use cases: Cut per-turn model spend in a Hermes agent; Add cheaper lane subagents to Claude Code; Screen web search results for hidden instructions · GitHub
Open Steps
A set of plain-language agent skills for Claude Code, Codex CLI, Cursor CLI and Gemini CLI that turn agent work into honest reports, clear verdicts and steps a non-engineer can follow.
4 use cases: Get a one-screen done-or-not report after each session; Turn agent requests into numbered steps you can follow; Check another agent's "done" claims before accepting them · GitHub
CLI Faq Shortcuts
A local Python history extractor and Agent Skill that finds the prompts you repeat in Claude Code, Codex and Cursor and turns them into reusable slash commands.
3 use cases: Turn repeated agent prompts into slash commands; Count and export your repeated prompts; Shortcut a repeated status check for your team · GitHub
Scrawl
Scrawl is a screen recorder that packs a narrated, annotated screen capture into a bundle (contact sheet, keyframes, transcript, manifest) so an AI agent can read it.
4 use cases: Show an agent a bug by recording your screen and voice; Let an agent zoom into one moment of a recording; Queue a capture into an OpenCode session · GitHub
Is GPT Nerfed
A macOS menu bar app and Codex plugin that checks whether the model you selected in Codex is the one actually answering, and flags silent downgrades.
3 use cases: Catch a silent model downgrade in Codex; Probe a brand-new session before a long task; Stop tool use when a session shows a mismatch · GitHub
Agent Py
Agent Py is a personal AIOps project that investigates alerts with tool-backed evidence, retrieves runbooks, and reviews remediation proposals.
4 use cases: Run the local alert investigation stack in one command; Check backend and frontend health before editing code; Run the synthetic incident evaluation against one API · GitHub
Droidline
Droidline is a tool that lets code, the command line, or an AI agent control real Android phones through a paired PC server, without ADB, a USB cable, or root.
5 use cases: Test your Android app on real phones from a script; Let an AI agent look at and operate a phone; Run one routine across a shelf of phones, each with its own proxy · GitHub
Chat UI
Chat_UI is a zero-dependency LLM chat client made of two files, under 100 KB, that runs in a web browser.
4 use cases: Chat with an LLM API from one browser page; Tune LLM generation parameters in a simple chat client; Save a chat session and resume it later · GitHub
Bricks
Bricks is an open-source component layer that gives coding agents implementation-aware contracts for building interface components that fit an existing project.
4 use cases: Build a project-native footer for a SaaS site; Design an agent task input with full states; Build an accessible FAQ section in an existing app · GitHub
Awesome Jev
An unofficial curated list of Jev use cases, SDKs, tools, and learning resources for TypeSafe AI's System One model, which returns typed decisions instead of generated text.
4 use cases: Route support tickets by intent and urgency in Python; Try Jev routing rules in the browser Playground; Add Jev decisions to a JavaScript or TypeScript app · GitHub
Insomnia
Insomnia is a macOS menu bar app that keeps a MacBook awake for timed sessions, including with the lid closed, and undoes its power and lid changes when a session ends.
4 use cases: Keep a MacBook awake for a timed agent run; Test lid-close behavior before a long unattended run; Recover a Mac after a session that could not be undone · GitHub
Triton Ascend Agent Dev Kit
A Shell toolkit of guides, project instructions, and 24 task-specific Agent Skills that helps AI agents write Triton-Ascend kernels for the Ascend NPU.
3 use cases: Set up an agent workspace for Triton-Ascend work; Have an agent draft a Triton-Ascend kernel; Delegate a Triton-Ascend task to a matching skill · GitHub
Herdr Auto Title
Herdr plugin that automatically names Herdr tabs and panes after the work in them, using the current directory, Git branch, terminal activity, and Claude Code sessions.
4 use cases: Auto-name Herdr tabs from their Git branch and folder; Tell Claude Code panes apart in Herdr's goto panel; Show each workspace's active task in the Herdr sidebar · GitHub
Compact Adviser
compact-adviser is an agent plugin for Pi, Claude Code, Codex CLI, and Grok Build that uses the TypeSafe Jev model to judge whether a session is at a safe point to run /compact.
4 use cases: Get safe /compact hints or automatic compaction in Pi; Add /compact advice and optional auto-compact to Claude Code; Get hint-only /compact advice in Codex CLI on macOS or Linux · GitHub
Perch
Perch is a JavaScript command-line tool that lints code for semantic problems such as logic bugs, using decision models.
5 use cases: Find logic bugs in a codebase, worst first; Confirm a fix works without recording it; Write a custom rule for your own code standard · GitHub
Chums
Chums is a free, open source set of apps and a hub that lets you talk to Claude Code and Codex agents running on your Mac from an Apple Watch and an iPhone.
4 use cases: Talk to a Claude Code agent from your Apple Watch; Approve or deny agent actions from your wrist; Run a Codex agent next to your Claude Code agents · GitHub
Unlazy
unlazy is an anti-laziness skill for AI agents that makes them write a checkable acceptance ledger, run runnable gates, and report only what the evidence supports.
3 use cases: Gate a solo refactor with a verified GATES.md ledger; Split a large task into parallel agent leaves with scoped claims; Block Claude Code from ending a session with unmet gates · GitHub
Jev Mem
Jev-Mem is a Python research codebase for agent memory, where a System-One controller manages memory writes and graph-based retrieval and a System-Two language model writes the final answer.
4 use cases: Run the offline memory demo in under 15 minutes; Build and query Jev Mem over your own observations; Run Jev Mem memory decisions on a local Laya model · GitHub
Aside Codemode
aside-codemode is a JavaScript code-mode tool for Aside that runs ripgrep-backed search, file tools and batched browsing inside one sandboxed call so only the filtered answer reaches the model.
4 use cases: Find every file that mentions a term in one call; Turn many web pages into typed rows in one call; Count matches in a huge tree without flooding output · GitHub
Lithos Metal
lithos-metal is an open-source LLM inference engine that generates Metal megakernels and serves local models on Apple silicon Macs.
4 use cases: Run a local model as the backend for a coding agent; Call the local model from scripts over HTTP; Compare speculative and target-only generation · GitHub
DSH Plugin Shop
A plugin shop for DeepSeek Harness that lets you browse, install, enable, disable, and update dsh plugins from a catalog that is rebuilt daily and tracked in Git.
4 use cases: Install the plugin shop and browse DSH plugins; Have an agent install the shop and verify it loads; List your DSH plugin in the community catalog · GitHub
Lody
Lody is a shared workspace where a team can connect machines, run coding agents such as Claude Code and Codex on them, and share the resulting conversations and code changes.
4 use cases: Connect a machine and dispatch agent work to it; Share a live agent conversation with teammates; Have a main agent delegate bug work to parallel chats · GitHub
Awesome Jev Projects
Awesome Jev is a source-backed catalog of open-source projects that use Jev, a System-1 typed decision model, for agent decisions such as browser, desktop, and mobile control.
4 use cases: Find Jev agent projects that match a task; Query Jev projects by domain from a terminal; Vet browser and desktop control Jev projects safely · GitHub
Endo
Endo is a desktop chat app and harness that runs agents from Codex, Claude Code, Antigravity and Grok, and composes them into workflows and schedules.
4 use cases: Get a daily priorities summary on a schedule; Compare answers from Codex and Claude Code in one thread; Fan a question out to several agents with best-of-n · GitHub
Coucou
Coucou is an open source Mac notch app (with Windows, Linux and Android parts) that shows your AI coding agent sessions, lets you approve their permission requests, and adds voice dictation, call notes and an iPhone companion.
5 use cases: Approve Claude Code permissions from the notch; Track coding agents and approve them from iPhone; Dictate text into any Mac app with a held key · GitHub
Raw CLI
Raw is an open-source TypeScript agent harness with a working tool loop, context compaction, sessions, approvals, a terminal REPL, a local browser dashboard and ACP, where every tool, skill, hook and policy is an editable file.
4 use cases: Run a coding agent on your own model; Fork a built-in tool to change how it works; Package an agent so teammates can install it · GitHub
DSH Jev Plugin
A community plugin for the native DeepSeek Harness (DSH) that uses Jev or Luna Decisions as a judgment model for skill selection, task checks, log trimming, approvals and stage navigation.
4 use cases: Check a DSH turn against its Todo before it stops; Narrow a DSH skill catalog to the skills a task needs; Shorten long command logs and passing test output for the model · GitHub
Agent Jev
AgentJev-0.6B is a Python model server that scores a state, such as a diff, log, or trace, against questions you supply and returns calibrated probabilities in one forward pass without generating text.
4 use cases: Gate agent merges on test output probability; Pick the next tool call from a fixed option list; Score tickets or traces on your own rubric · GitHub
Lyapunov
Lyapunov is an LLM-first, 3D-native robotics and agent desktop framework that brings scene building, robot preparation, controlled simulation and camera data into one Linux workspace.
4 use cases: Prepare and run a robot action in simulation; Collect RGB-D camera data from a simulated scene; Build a collision-ready 3D scene from a GLB file · GitHub
DSH Commandcode Provider
A DeepSeek Harness provider plugin that adds Command Code model access, a live model catalog, and a settings page to dsh.
5 use cases: Run Command Code models in the dsh web app; Use Command Code models in the dsh terminal UI; Keep dsh sessions running across several Command Code accounts · GitHub
Clodfarm
clodfarm is a Docker-based farm of Claude Code agents that runs around the clock on your Claude subscription, splits missions into sub-agents, and paces itself on real usage limits.
5 use cases: Run coding tasks on a repo overnight, merged only if tests pass; Assign work to the farm from your phone; Triage new issues on a schedule · GitHub
Ciele
Ciele is a self-hostable TypeScript platform for building embeddable AI chat assistants that answer from your own websites, files, and FAQs, with escalation to human help desks and conversation analytics.
4 use cases: Run a private AI assistant on your own machine; Build a support assistant from your website and FAQs; Audit assistant answers from conversation transcripts · GitHub
Skills
A set of editable agent skills for planning, aligning, building, and debugging software with coding agents, taken from the author's .agents directory.
5 use cases: Align an agent on a change before any code is written; Split a plan into tracer-bullet tickets with blockers; Fix a bug with a failing test before the fix · GitHub
Task State Guard
TaskStateGuard is a Python library and CLI that reconciles stuck AI-agent task and delivery states in a local SQLite ledger.
3 use cases: Preview restart cleanup of stuck tasks before applying it; Record task and delivery closures only from real outcomes; Check a workflow ledger's health before trusting its history · GitHub
Kotlin Footguns
A set of agent skills and a scanner that flag 228 known failure modes in Kotlin, Compose Multiplatform and desktop JVM code, drawn from a production music app.
4 use cases: Scan a Kotlin change for known footguns before merge; Review a Compose screen against UI and state traps; Audit desktop packaging, signing and CI config · GitHub
Memorax Code
MemoraX Code is a memory plugin for AI coding agents that carries engineering lessons, repository knowledge, and working preferences into later sessions.
4 use cases: Resume a coding task with lessons from earlier sessions; Onboard to a repository with Repo Memory; Carry your preferences and procedures into new sessions · GitHub
Better Claude
A set of plugins and mods by midudev that add small features to Claude Code, such as image previews, usage meters, commit helpers and spoken turn summaries.
4 use cases: Split your changes into clean Conventional Commits; Hear what Claude did after each turn; Watch your 5-hour and weekly usage at a glance · GitHub
TV Home
tv-home is a set of scripts and a small Android home screen launcher that debloat a slow Android or Google TV over Wi-Fi and replace its home screen.
4 use cases: Debloat a slow Android TV over Wi-Fi; Install a fast Apple TV-style home screen; Type text into the TV from your laptop · GitHub
Agentacct
agentacct is a local-first dashboard and terminal app that shows what your coding agents did, the tokens and estimated cost behind that work, and whether their claimed results were verified.
4 use cases: See what your coding agents did and what it cost; Check whether an agent's 'done' claim is verified; Review token use and cost by agent and model · GitHub
Golive Skill
GoLive is an open-source Agent Skill and zero-dependency Node CLI that takes an agent-built app live on the user's own hosting, database, domain, email and payment accounts, with approval steps before any change.
4 use cases: Take an agent-built app live on your own accounts; Tear down a test launch and list what remains; Connect a custom domain and transactional email · GitHub
Open Agent Core
OpenAgentCore is a self-hosted, open source Agents API that runs Codex, Claude Code or MiniMax Code inside sandboxes, using the official OpenAI SDK.
3 use cases: Run a coding agent Session through the OpenAI SDK; Point an existing OpenAI SDK app at self-hosted Core; Run agent Sessions on your own Linux, macOS or Windows machine · GitHub
Anti Slop
antislop is a set of rules and skills that stop an AI coding agent from shipping generic UI, filler copy, and AI-shaped code.
4 use cases: Build a landing page free of generic AI UI; Remove filler and fake stats from marketing copy; Audit existing code and remove AI-slop comments · GitHub
Iphone Duo Skills
A set of agent skills, installed as SKILL.md files, that give AI coding agents structured guidance for building and porting apps on iPhone Duo, Apple's first foldable iPhone.
5 use cases: Audit a SwiftUI or UIKit app for iPhone Duo readiness; Make a capture app handle iPhone Duo's two front cameras; Review a screen design for foldable poses and fold avoidance · GitHub
Esf
ESF is a self-hosted, open source software factory that runs coding agents in isolated CubeSandbox microVMs, verifies their changes, and preserves the patch and execution evidence.
4 use cases: Turn a coding task into a verified patch; Pause agent runs for human approval; Submit a reviewed run from a manifest · GitHub
Codex ChatGPT Web
A launcher that lets you use ChatGPT Web models, including Pro, as a native model inside Codex, with optional MCP tools for local files and terminal access.
4 use cases: Use ChatGPT Web models inside Codex without Codex quota; Let ChatGPT use files and terminal in a Codex task; Send prepared prompts to ChatGPT with Zero Risk mode · GitHub
Jev Lint
jev-lint is a command-line lint tool that uses an AI classifier to flag code, commits, shell scripts and documents whose claims contradict what they actually do.
5 use cases: Find doc comments that contradict their TypeScript code; Review only the lines a pull request changed; Check commit messages and staged changes against AGENTS.md · GitHub
Laya MLX
laya-mlx is a Python runtime that runs Laya typed decision models natively on Apple Silicon using MLX, with no text generation, PyTorch or cloud API.
4 use cases: Route support tickets to a team on your Mac; Answer several typed questions about one message; Watch a local Snake agent make live decisions · GitHub
Context Mode
Context Mode is an MCP server and hook set for AI coding agents that keeps raw tool output out of the context window, persists session memory in SQLite, and routes work across many clients.
5 use cases: Keep large tool output out of Claude Code's context; Count or summarize code with one script, not many reads; Pick up a coding task after the chat compacts · GitHub
Monid
Monid is a TypeScript connector layer that describes AI tool providers and their endpoints in one declarative format, so agents can call many APIs through one interface.
4 use cases: Add your API to Monid as a connector; Run a live web search endpoint from the terminal; Pick the right search endpoint from the catalog · GitHub
Project Maya
Project Maya is a free, local engine and server that runs large GLM-5.3-Flash and DeepSeek-V4.1-Flash-1017 models privately on your own NVIDIA or AMD GPUs.
3 use cases: Run a private coding-agent model on your own GPU; Ask questions about confidential files without cloud upload; Run DeepSeek-V4.1-Flash locally with picture input · GitHub
Prismantis
prismantis is a TypeScript mod for Claude Code that restyles replies with themed colors, tables, code blocks, diagrams, charts, figures, optional LaTeX math, copy buttons and tool rows, and it also runs in Codex.
5 use cases: Install Prismantis and pick a theme for Claude Code; Draw a flowchart or chart inside a Claude Code reply; Get a timeline or before-and-after picture in a reply · GitHub
Session Knowledge
session-knowledge is a local, stdlib-only Python tool that indexes Claude Code session transcripts into a searchable knowledge base and exposes 13 MCP tools for searching them.
5 use cases: Find existing code for a service before writing it again; Recover the reasoning behind a past decision; Look up the commands that failed last time · GitHub
Foremerge
Foremerge is an open-source Rust coordination protocol that lets coding agents announce planned changes, see overlapping plans, and share claims and verification results above Git.
4 use cases: Catch conflicting agent plans before code is written; Set up Foremerge for a repo's coding agents; Claim a code area so other agents see ownership · GitHub
GPT with Taste
A set of behavioral instructions and event hooks that steer GPT models in OpenAI Codex, Claude Code, and other coding harnesses.
5 use cases: Install GPT with Taste hooks into Codex; Add the recipe to Codex with prompt files only; Have a coding assistant install and check GPT with Taste · GitHub
Pooled
Pooled runs one open language model across the browsers and devices in a room, so people can chat with it or use it from coding agents.
4 use cases: Chat with a big open model Pooled across your devices; Point a coding agent at a Pooled room's model; Have a room's model build and fix a small app · GitHub
Web GPT
WebGPT is a Codex skill that hands tasks off to ChatGPT on the web through a local Worker, so the work does not use up Codex quota.
3 use cases: Offload a research task to ChatGPT Pro from Codex; Build planned project features in parallel with subagents; Open ChatGPT with terminal access to a project · GitHub
Jev Review
Jev Review is a local MCP server that gives AI coding agents structured, per-dimension quality scores for their code changes, powered by the Jev API.
3 use cases: Score Claude Code changes for quality during a task; Score Codex changes against quality dimensions; Refactor in scored steps until deltas plateau · GitHub
Agent Router
Agent Router is a TypeScript command-line tool that picks an AI coding agent (Cursor, Claude Code, Codex, or OpenCode), a model, and a reasoning effort for a task, then starts that agent in a Herdr pane.
4 use cases: Preview which agent and model a task would use; Launch a task in a Herdr pane with the chosen agent; Check subscription quota before routing a task · GitHub
Openbot
OpenBot is a local-first desktop app and optional always-on server for running persistent AI coding and assistant agents, each with its own workspace, queue, and conversation.
4 use cases: Run persistent agents on a headless server; Split a coding task across Codex and Claude agents; Have an agent research and test sites in its own browser · GitHub
No Human
no_human is a free, open-source Python tool that turns a ticket into a reviewed pull request on your own machine, with a plan, an independent model review, and a tamper guard.
4 use cases: Turn one issue into a reviewed pull request; Stop AI changes that delete or weaken tests; Check that a Python bug fix really fixes the bug · GitHub
Sentry
Sentry is an external runtime failure management layer for LLM agents that detects failures, guides recovery, and learns reusable lessons from recoveries that work.
3 use cases: Add runtime failure recovery to an agent loop; Benchmark agent recovery on AppWorld tasks; Summarize Sentry's recovery findings from the paper · GitHub
Nuvex Services
A TypeScript set of off-chain services for Nuvex: a read-only HTTP API, a chain indexer that writes PostgreSQL or a JSON snapshot, price-provider adapters, and an empty AI directory.
3 use cases: Serve Nuvex network counts from an indexed Postgres store; Confirm the services repo passes format, lint, tests, and schema checks; Write indexer output to a JSON snapshot instead of Postgres · GitHub
Speech CPP
speech.cpp is a C++ command and library on ggml for speech synthesis, speech recognition and voice activity detection, running on the CPU, Metal or Vulkan with an OpenAI-compatible server.
5 use cases: Transcribe long Japanese recordings with timestamps; Caption the microphone live as you speak; Generate spoken narration audio from a text file · GitHub
Superpowers
Superpowers is a set of composable skills and instructions that make coding agents follow a structured software development method: design first, then a written plan, test-driven implementation, review, and finishing the branch.
4 use cases: Build a feature from design to reviewed code; Find the root cause of a bug before fixing it; Isolate agent work on a branch and finish it cleanly · GitHub
Okf Agent Memory
A Go command-line tool and library that keeps AI agent project memory as a Git-native OKF v0.2 knowledge bundle, with validation, search, and an embedded MCP server.
4 use cases: Validate a project knowledge bundle before each commit; Let a coding agent search project decisions before writing code; Look up rules across project, vendor, user, and system scopes · GitHub
One Jev
OneJev is a Python package and set of Qwen-based multimodal models that return calibrated probabilities for typed questions about screenshots, photos, video and text in one forward pass.
4 use cases: Pick the next UI action from a screenshot; Run the One Jev API and playground in Docker; Run a small One Jev model on a laptop with llama.cpp · GitHub
Omnirush Gui
A desktop coding agent app for working on your own projects, using models from your omnirush.ai account.
3 use cases: Run a coding agent on a local project from the desktop app; Have the agent branch, commit and open a pull request; Let the agent query local files and SQLite through MCP · GitHub
Esp32 Iphone Agent
An ESP32-S3 board that acts as an AirPlay screen mirror and USB mouse and keyboard, paired with an iPhone app that lets a vision model see and operate the phone.
5 use cases: Have an AI model complete a task on a real iPhone; Replay and export phone agent runs for analysis; Start saved phone tasks from Siri or Shortcuts · GitHub
Opengrok
opengrok is a config sidecar that lets an existing Grok Bot install use other AI models through a model picker, provider wire maps, and a health doctor.
4 use cases: Bind non-Grok models to Grok Bot agents; Swap models mid-chat from the LiquidGlass panel; Keep model bindings working after Grok Bot updates · GitHub
Openpactprotocol
PACT is an open TypeScript protocol, with schemas, a client library, and reference code, that lets personal agents call businesses' agents with signed identity and customer-granted permissions.
4 use cases: Run PACT conformance tests against a Provider; Build a personal agent that calls business agents; Validate PACT messages with protocol schemas · GitHub
Codexqa
CodexQA is an open-source pack of local Agent Skills for Cursor, Claude Code, Codex, and OpenClaw that verifies code quality, change impact, defects, requirements, test cases, and browser flows before AI-generated code is merged.
5 use cases: Review a branch before merge and get a shareable HTML report; Scan a branch for bugs and security issues before merging; Map a diff's impact and generate tests for uncovered paths · GitHub
Openqodex
OpenQodex is an open source tool that reviews your code change before you push, combining scanners with a separate AI reviewer that uses your own Claude Code or Codex login.
4 use cases: Try OpenQodex on a demo repo with planted bugs; Review your change before you push it; Review a pull request on your machine before merging · GitHub
Pi Bolt
Pi-Bolt is a compiled-ahead-of-time build of the Pi coding agent that starts faster and uses less CPU than Pi on Bun or Node.
4 use cases: Start the Pi coding agent faster on your machine; Compile your daily Pi extensions into Pi Bolt; Run Pi Bolt in your editor's agent panel · GitHub
Master Agent
A C++17 local-first agent runtime and sparx CLI that matches deterministic phrases first, calls a model only when configured, and runs only validated tool calls.
5 use cases: Run the offline agent demo in under an hour; Add phrase-routed tools to a C++ app; Stop repeated tool runs after a crash · GitHub
Open Vetta
Open Vetta is an open-source, local-first desktop AI agent that works on your files, tools, and models (bring your own key) and can be extended with skills, MCP servers, plugins, and themes.
5 use cases: Run a coding task on a local project with your own model; Build and hot-reload a Vetta plugin outside the source repo; Repeat a proven task across many directories on a schedule · GitHub
Oriveo
Oriveo Community Edition is an open-source, bring-your-own-key AI chat client for iOS, Android, and the web that talks directly to model providers and local model servers using your own keys.
5 use cases: Run a private chat app in your browser with your own key; Chat with a local model server through a relay endpoint; Build and run the iOS app on your own iPhone · GitHub
Cortico
Cortico is a free, open source TypeScript agent harness for building persistent, always-running AI bots that handle mixed real-time input from chat platforms, live games and other environments.
5 use cases: Run a persona bot locally and chat with it in the WebUI; Build a Cortico extension by talking to a coding agent; Run an AI VTuber with a Live2D model, streaming TTS and OBS overlay · GitHub
Claude Share
Shared Sessions is a Claude Code plugin that lets a host share a live session link so teammates can join, see the same prompts, replies and tool results, and work together.
4 use cases: Pair with a teammate on a live Claude Code task; Review Claude's replies and tool results together; Let guests watch or ask before edits and commands · GitHub
Time to First Token
A 10-week self-study roadmap for running LLM inference serving in production, built around one OpenAI-compatible service that gets deployed, instrumented, load tested, optimized, and benchmarked.
4 use cases: Serve a 7-8B model behind an OpenAI-compatible endpoint; Chart TTFT and queue depth on a live LLM service; Compare vLLM and SGLang on a shared-prefix workload · GitHub
Debroid
Debroid is a headless Android debugger CLI that lets AI agents set breakpoints, catch exceptions, inspect and change variables, and step through a live Android app over JDWP.
3 use cases: Catch an app crash and read live state at the throw; Pause at a suspect line and inspect its variables; Find which code writes a field that goes wrong · GitHub
Herdr Gpui
Herdr UI is a native Rust and GPUI desktop client for macOS, Linux, and Windows that shows the terminal sessions, workspaces, Git worktrees, and agent activity of a Herdr daemon you installed yourself.
4 use cases: Install the native Herdr client and open your sessions; Review an agent's changes and send notes back to it; Mark up a page an agent opened and send notes back · GitHub
Halogen Flash Server
halogen-flash-server is a Shell-based inference server that runs the Qwen3.8-Flash-Next model on AMD Strix Halo (gfx1151) GPUs through an OpenAI-compatible API, packaged as a container image.
5 use cases: Run a local Codex agent on a Strix Halo box; Point Claude Code at a local halogen server; Resume a long agent session after a server restart · GitHub
SPACE
SPACE is a Rust sandbox platform that runs code in isolated Firecracker microVMs, controlled through an HTTP API.
5 use cases: Run agent-written code in an isolated sandbox; Point existing E2B Python SDK code at SPACE; Build the SPACE guide and API docs on your machine · GitHub
DSH Auto Review
dsh-auto-review is a DeepSeek Harness plugin that sends approval requests to a read-only second-model reviewer, which returns allow or deny verdicts with reasons and fails closed by default.
3 use cases: Let a second model approve routine bash and write requests; Trace why an approval was allowed or denied; Block risky commands with rules and a stricter fallback · GitHub
Jevcore
jevcore is a TypeScript package that gives DeepSeek Harness, MCP hosts, and plain Node scripts access to TypeSafe Jev, which returns typed yes/no, choice, and score judgments with calibrated probabilities, offline by default.
4 use cases: Route support tickets with typed Jev answers; Install the jev plugin into DeepSeek Harness from a checkout; Serve the three Jev tools to a non-DSH MCP host · GitHub
Pgbot
pgbot is a read-only Go command-line tool that reads PostgreSQL's own statistics views and prints a findings-first health report with changes since the last run.
5 use cases: Triage a slow PostgreSQL database in ten seconds; Find the queries that eat the most database time; Spot unused indexes and likely missing indexes · GitHub
Artifact Server
Artifact Server is an open source, self-hostable place to publish, review, comment on, version, and share HTML, images, and video artifacts, with a built-in MCP server for agents.
4 use cases: Publish an agent's design doc for full-screen review; Run a private artifact server on your laptop; Host a private artifact hub for your product team · GitHub
Gap Trap
gap-trap is a skill for Claude Code and Codex that reads a repository, writes contracts and gates so AI-written code follows the codebase's rules, and adds CI checks that prove the tests can fail.
4 use cases: Set up gates so agent code stays correct; Find existing rule violations before they spread; Refine rules after incidents and keep gates current · GitHub
Medula
medula is a headless devtools package that lets a coding agent inspect and change the internal state of a running web app through MCP.
3 use cases: Verify a bug fix by reading live component state; Test cart edge cases by editing live state; Inspect Vue Pinia stores and walk routes with an agent · GitHub
Roamgate
Roamgate is a browser and PWA client for Herdr that lets you view and control workspaces, coding agent sessions, terminals, files, and diffs from desktop or mobile.
4 use cases: Continue a Herdr workspace on your phone; Review agent code changes and diffs in a browser; Monitor agent sessions and run tests from a phone · GitHub
Janas
Janas is a family of C tools whose first component, Janas-LLM, runs large mixture-of-experts language models on ordinary computers, with or without a GPU.
5 use cases: Chat with an 80B model on a 32 GB laptop; Measure your own machine's speed with janas-bench; Chat with a local model on a Windows laptop · GitHub
Pullboard
Pullboard is a local coordination board, stored in a git repo, that lets AI agents build software from specs you approve, with a second agent verifying each change before it counts.
5 use cases: See a builder and verifier agent work one rule; Set up agent coordination in an existing repo; Ship one approved feature with a second-agent check · GitHub
Awesome Remote MCP Servers
A curated list of remote Model Context Protocol servers, which are hosted endpoints you connect to by URL instead of installing locally.
4 use cases: Connect a hosted Linear MCP server to Claude Code; Add a remote MCP server to Cursor with mcp.json; Find a hosted MCP server for a specific task · GitHub
Dscode
DSCODE is a terminal coding agent for macOS and Linux, built on DeepSeek Harness, with a persistent shell, sessions that other sessions can message, auto approval and session telemetry.
4 use cases: Review a Git diff in a script or CI job; Hand a code review to another DSCODE session; Get an independent review before committing a change · GitHub
Pulse
Pulse is a free, open-source macOS app that shows how much of each AI coding tool's usage limit you have left, as rings on the edge of the screen.
4 use cases: See remaining limits for all your AI coding tools; Get warned before a rate limit stops your coding; Review AI token spend across local clients by day · GitHub
Qwen Audio Agent
qwen-audio-agent is a JavaScript realtime voice runtime that lets people talk with an AI Agent while the Agent runs tool calls and background tasks in the same conversation.
4 use cases: Delegate coding tasks to an Agent by voice; Keep a floating voice assistant on your desktop; Talk to an Agent in the browser with live camera frames · GitHub
Mu
mu is a command line and desktop coding agent that sends routine permission and context decisions to a small judge model so the main model can focus on the code.
3 use cases: Run a coding session that asks only about risky steps; Compare a judge in shadow mode before trusting it; Code with the Mu desktop app and no terminal · GitHub
Vibex
vibeX is a tool that generates architecture diagrams and checkable docs from your Prisma schema, OpenAPI document, GraphQL SDL, or source code, delivered as one standalone HTML file.
5 use cases: Draw the real data model from a Prisma schema; Map every API endpoint from an OpenAPI document; Document a service and fail CI when the docs drift · GitHub
Janus
Janus is a small macOS menu bar app that switches between saved Claude Code and Codex accounts and clears developer caches by moving them to the Trash.
4 use cases: Switch between Claude Code accounts in one click; Check which Claude account has room left before work; Switch between Codex accounts and API-key sign-ins · GitHub
Awesome REA
A curated list of tools, case studies and research papers for reverse engineering websites, games, apps, binaries, firmware and communication protocols.
4 use cases: Extract a website's design system from its live pages; Make minified JavaScript readable before auditing it; Turn captured app traffic into an API spec · GitHub
Openreality
Open Reality is an open-source toolkit that turns phone video into AI-queryable 3D scenes and exposes them to AI assistants through MCP tools.
5 use cases: Scan a room from phone video and query it; Calibrate a scan and measure distances in it; Plan a walking route through a scanned space · GitHub
ML View
MLView is an experimental VS Code viewer and skill that lets a coding assistant turn an ML workflow into an interactive, source-linked diagram.
3 use cases: Turn a training config into a source-linked diagram; Check each diagram claim against its cited lines; Update a workflow diagram after the code changes · GitHub
Reason Network
Reason Network is a peer-to-peer daemon that lets your team's coding agent sessions see each other and exchange messages and small text files.
4 use cases: Ask a teammate's coding agent a question directly; Approve a teammate's request through a pop-up; Send a small text file to a teammate's agent · GitHub
Ripwire
ripwire is a zero-dependency C++23 command-line tool and optional MCP server that gives coding agents a ranked map of a repository and checks what they changed.
4 use cases: Find where to change code without reading the repo; Check what an edit breaks before calling it done; Find the real callers before changing a behavior · GitHub
Rizzo Flow
Rizzo Flow is an open, local-first tool that returns typed, probabilistic decisions (yes/no, choices, rubric scores, numbers) from unstructured state using a local LLM over an HTTP API.
4 use cases: Get typed LLM decisions from text on your own machine; Triage a support ticket from the command line; Run existing TypeSafe Jev API code against localhost · GitHub
Unbiased Strata
UnbiasedStrata is a C++ fork of PolyStrata that builds two local inference engines and a front end, unbiased-strata, which serves an OpenAI-style chat API for one NVIDIA GPU on Linux.
4 use cases: Serve a local chat API on a machine with no network; Ask a local model questions about screenshots; Test the server front end without a GPU or model · GitHub
AI Engineering From Scratch
A free, open source course of 542 AI engineering lessons with runnable code in Python, TypeScript, Rust, and Julia, plus three graded projects.
4 use cases: Learn AI engineering lesson by lesson in your agent; Catch a retrieval ranking regression before release; Separate child work from parent time in an agent trace · GitHub
Quackd
quackd is a Python command line that connects robots you own, gives each one an LLM as its pilot, and lets you command one robot or a flock with a goal.
5 use cases: Try a duck goal in the simulator with no robot; Command a real SO-101 arm with a plain-English goal; Rehearse arm task files in simulation before the arm · GitHub
Rome
Rome is a self-hostable agent operating system where humans and agents build, reuse, and refine actions, skills, and apps over time.
4 use cases: Run a code review loop until no merge blockers remain; Build a missing app from a plain-language description; Track a product price and get alerted on a drop · GitHub
Grok App
Grok App is an open-source desktop workbench that gives a GUI to the local Grok Build CLI (grok agent stdio), covering sessions, projects, files, media, and automations.
5 use cases: Run a local Grok agent on a project from a desktop app; Run parallel agent tasks in Git worktrees and review diffs; Check and steer your desktop agent from a phone · GitHub
Think Term
ThinkTerm is an open-source terminal with a built-in multiplexer, written in Rust, whose mux server keeps shells and coding agents running after the window closes.
4 use cases: Keep long builds and agents running after you close the window; Let a coding agent read and drive other terminal panes; Watch agents on several machines from one Think Term sidebar · GitHub
Ruflo
Ruflo is an agent meta-harness for Claude Code and Codex that adds swarms, memory, hooks, MCP tools and a console around the coding agent.
5 use cases: Connect Codex to Ruflo's MCP server; Watch agent runs in a terminal console; Add the Ruflo console to Claude Code · GitHub
DeepSeek Harness Handbook
A community-maintained handbook of source-backed guides for understanding, running, debugging, and extending DeepSeek Harness.
4 use cases: Find the right fix for a failing Harness run; Score a Harness setup against a reference scorecard; Vet an ecosystem plugin before you install it · GitHub
Codex Bridge
A Claude Code plugin that adds image generation (gpt-image-2) and GPT-5 subagents by calling the Codex CLI through your existing ChatGPT login, with no OpenAI API key.
4 use cases: Generate a project image from a text prompt; Build a matching favicon and OG card set; Delegate a repo-wide rename to GPT-5 · GitHub
Agentenv Framework
AgentEnv Framework is a Python SDK and CLI for building, deploying and running containerized agentic environments and the tasks that grade agents inside them.
4 use cases: Run the hello task to check your install; Run local environments in the explorer; Build a custom environment with the SDK · GitHub
Hatoba
Hatoba is an open-source desktop SSH client with an AI assistant and end-to-end encrypted sync stored in your own Cloudflare account.
4 use cases: Build the Hatoba desktop app from source; Sync SSH hosts and keys across devices, encrypted; Watch every saved server's health on one page · GitHub
Seiso
seiso is a Rust command-line linter that checks Markdown project docs against a shared convention so that humans and AI agents can rely on them.
3 use cases: Check project docs against a shared Markdown convention; Have an agent repair Seiso findings in project docs; Gate doc changes with Seiso in Claude Code hooks or CI · GitHub
Claude Flightdeck
Flightdeck is a Claude Code mod that adds a live dashboard to your terminal showing context, cost, permission checks, advisor consults and subagents.
4 use cases: Watch context, cost and rate limits during a session; Audit every permission check a session makes; Follow subagents as cards or swimlanes · GitHub
Codex Command Center
A Windows-first local desktop app that organizes Codex projects, engineering tasks, sessions, Git changes, Skills, and MCP configuration.
4 use cases: Resume a Codex engineering task from where you left off; Review Codex file changes and commit them in focused batches; Build the desktop app from source and run native checks · GitHub
Sloptrim
Sloptrim is a local command-line tool and agent plugin that scores prose files for AI-writing patterns, using only the Python standard library with no network and no model.
4 use cases: Score a prose draft for AI-writing tells; Guard prose an agent saves in Claude Code; Add a prose contract to AGENTS.md and Cursor rules · GitHub
Lint
A linter for Tailwind design systems that lets you define allowed styling and gives coding agents fix suggestions drawn from your own components and theme.
3 use cases: Block arbitrary Tailwind values in agent-written UI; Lock component styling with per-component contracts; Make coding agents fix Lint errors before finishing · GitHub
Synapass
Synapass is an open-source OpenAI-compatible gateway and control plane that routes each AI request to a provider under policy, handles failover, caching, budgets, and request tracing.
4 use cases: Send all app AI calls through one OpenAI-style endpoint; Trace which provider and policy handled a bad AI answer; Set spending limits so one team cannot run up the AI bill · GitHub
RSI Jev
RSI-Jev is a set of typed-decision models (yes/no, pick-one and rubric questions) that AI agents trained in a self-improving loop, with the checkpoints, code and failed experiments published.
5 use cases: Serve a decision model behind a local API; Make a typed decision in Python without a server; Run the 4B decision model on a Mac · GitHub
DSH Taskboard
DSH Taskboard is a native, local SQLite-backed task manager plugin for DeepSeek Harness, with a Web UI, Agent tools, a JSON CLI, and a Skill.
3 use cases: Install Taskboard into a running DeepSeek Harness profile; Have an agent submit finished work to in_review for human sign-off; Track local project tasks in a native Web UI board · GitHub
Nurb
nurb is an agentic CAD tool that lets an AI model 3D printable parts, check them for print problems, and show them in a live viewer.
5 use cases: Design a printable hose adapter from a text description; Catch print problems before the printer runs; Estimate print time and filament cost before printing · GitHub
Hyperfide
Hyperfide is a local agent workflow CLI that indexes a code repository into a graph, runs planned task loops, and keeps decisions and release notes for humans and AI agents.
6 use cases: Turn a feature idea into a verified task plan; Give an agent a compact brief of a code area; Fan a task wave out to parallel Codex or Claude Code sessions · GitHub
Paxeer X Network
Paxeer X Network is a monorepo holding the Paxeer X EVM chain node (paxd, Go) and the LayerX kernel (layerxd, C17), with SDKs, an MCP server and a bridge relayer for agent and payment workloads.
4 use cases: Build the LayerX kernel and run its test gate; Build and test the paxd Paxeer X chain node; Connect a wallet or app to the Paxeer X chain · GitHub
Setup Gateway
A zero-dependency Node.js CLI that points Claude Code, OpenCode, and Codex CLI at any OpenAI-compatible gateway, with automatic backups and a restore command.
4 use cases: Point Claude Code at your own OpenAI-compatible gateway; Set up Codex CLI with gateway models in the /model picker; Restore the last working config after a bad gateway change · GitHub
Sno Station Skills
Sno Station Skills is a set of 31 open source skills that install into Claude Code and Codex so they can review each other's work, hand off tasks when quota runs out, report each morning, delegate with written success checks, and learn from past sessions.
5 use cases: Get a cold second review from the other vendor's agent; Keep an overnight build going when one agent runs out of quota; Read a one-page brief when you come back to the desk · GitHub
Sno Station
Sno Station is an open source tool that runs on your computer and lets Codex, Claude Code, Cursor, Hermes Agent and OpenClaw share one encrypted memory, hand work to each other, review each other's work, and run an overnight Self Learning step.
4 use cases: Hand off a half-finished task when one agent hits its limit; Get a code review from a different agent; Review the skill changes Self Learning proposes each morning · GitHub
Agenttrail
Agenttrail is a local observability tool that shows the project structure, file activity, tasks and role contributions of AI coding agents such as Codex and Claude Code.
4 use cases: Watch coding agents work live in a 3D kitchen; Preview the kitchen interface with scripted example steps; Map project structure and component progress from PLAN.md · GitHub
Maid Kit
MaidKit is a cross-platform Flutter app for managing servers over SSH, with an optional outbound-only MaidCafe daemon for fleet monitoring and alarms.
4 use cases: Run Maid Kit locally to manage servers over SSH; Build a Linux AppImage of Maid Kit for a workstation; Monitor a server fleet with the MaidCafe daemon · GitHub
Vibe Coding Toolkit
A collection of documentation, prompts, and templates that set up a machine or a project for building with AI coding agents such as Claude Code and Codex.
4 use cases: Start a Next.js project from a product idea; Bring an existing project up to the toolkit setup; Prepare a development machine for agent-driven work · GitHub
Autoprompt Skill
Autoprompt is a coding-agent skill that reviews, fixes, and rechecks a coding task inside supported coding agents.
4 use cases: Fix a bug and add a regression test with Claude Code; Build a multi-part feature with parallel Codex work; Compare options against a codebase before deciding · GitHub
Clawd Conduit
A wiring layer that connects Claude Code's 15 lifecycle hook events to the Clawd on Desk desktop pet, so the pet reflects session state and permission prompts can be answered from the desktop.
4 use cases: Approve Claude Code permission prompts from the desktop; Get a system banner when Claude Code needs input; Watch Claude Code session state on a desktop pet · GitHub
Starling
Starling is an open source project that builds a Linux desktop environment with a Swift shell and compositor, a Swift port of the Flutter framework, and first-party apps such as a terminal and an office suite.
4 use cases: Build and run the Starling desktop from source; Package the Starling desktop as a Debian package; Build the Starling Terminal from source · GitHub
Tellurion
Tellurion is a free, open source macOS terminal (a fork of Ghostty) that shows live status, edits, diffs and approval prompts for several AI coding agents running across your projects.
4 use cases: Watch every coding agent from one screen; Undo an agent turn that broke your files; Hand off a stuck task to another agent · GitHub
Glyd
Glyd is a Rust library and CLI that keeps bf16 model weights and KV cache in fewer bits on the GPU, restoring exact values as the model runs, and also serves as a lossless file compressor.
4 use cases: Run Qwen3.5-9B on a 24 GB GPU with long context; Serve Qwen3.5-9B through vLLM with Glyd quantization; Load a bf16 model in Python with Glyd · GitHub
MCP Server
An MCP server that gives AI agents tools to order a private phone number in a chosen country, read the SMS verification code from it, and hand the number back.
4 use cases: Get an SMS verification code for an agent signup; Compare country coverage and delivery rates before ordering; Recover when an SMS code does not arrive · GitHub
Berth
Berth is a single Go binary that lets local AI engines, games and GPU apps take turns on one NVIDIA GPU instead of loading at once and crashing with out-of-memory errors.
4 use cases: Stop local AI engines from crashing each other on one GPU; Let games take the GPU while you play and resume AI after; Share the GPU between llama-swap and ComfyUI without ping-pong · GitHub
Fieldcraft
FieldCraft is an open-source, pure Rust finite element and CFD application that meshes parametric parts and solves structural, thermal and fluid problems, with a CLI and MCP server for AI agents.
3 use cases: Solve a ready-made bracket analysis in minutes; Have an AI agent mesh and solve a beam over MCP; Check thermal stress in a heated steel bracket · GitHub
Freellmapi
FreeLLMAPI is a self-hosted router that puts many providers' free LLM tiers behind one OpenAI-compatible /v1 endpoint, with failover and encrypted key storage.
5 use cases: Point an OpenAI SDK app at one free-tier endpoint; Run Claude Code against the free model pool; Add a local model server beside the free tiers · GitHub
Mobile Harness
Mobile Harness is a no-root Android app that runs an Ubuntu 20.04 Linux environment on your phone, where you can chat with coding agents, edit projects, run commands, and preview web apps.
5 use cases: Build and install an Android app entirely on your phone; Have Claude Code build a web app from your phone; Preview a local Vite or Express server on your phone · GitHub
Octop Harness
octop-harness is a Python library that wraps deepagents into a production-oriented agent runtime with model routing, memory, tools, multi-agent registry, storage backends, and a CLI.
5 use cases: Run a production agent with swappable model providers; Build an assistant that remembers users across sessions; Let an IDE or terminal AI drive your agent over ACP · GitHub
Claude Mem
Claude-Mem is a persistent memory system that captures what coding agents do in a session, compresses it with AI, and injects relevant context back into later sessions.
4 use cases: Carry Claude Code project context into new sessions; Find past decisions and bug fixes before you repeat them; Give Codex threads in T3 Code shared project memory · GitHub
Paperclip Guide
A companion guide that walks you through installing Paperclip, an open-source app for running a team of AI agents like a company, and building an AI IT department with it.
5 use cases: Build an AI IT department with Paperclip; Assign agent work with approvals and watchdog checks; Schedule a daily storage check and agent standup · GitHub
Open Intelligent UI
A TypeScript chat app that streams answers as interactive UI, such as maps, photos, charts and working tools, built on the OpenUI library.
4 use cases: Run a chat app that answers trip questions with maps and itineraries; Prototype data dashboards that answer questions with charts and tables; Build interactive money tools, such as a loan or bill-split calculator · GitHub
Openui
OpenUI is an open, framework-agnostic Generative UI framework that lets models stream structured UI, built around the compact OpenUI Lang and runtimes for React, Vue, Svelte, and Angular.
5 use cases: Scaffold a streaming generative UI chat app; Add streamed OpenUI rendering to a React app; Generate model prompts from a component list on the server · GitHub
System1 Agents
System1-Agents is a Python toolkit that gives AI agents System 1 decision models (Jev, Laya, Cua-S1) for browser, computer, game and robotics tasks.
4 use cases: Route support tickets to a queue in one call; Hand single decisions to a System 1 agent from Claude Code; Compare Jev and a chat model on the same agent · GitHub
Think Watch Lite
ThinkWatch Lite is a local gateway on macOS, Windows and Linux that sits between AI clients such as Claude Code and Codex and their upstreams, recording requests and filtering keys and tool calls.
4 use cases: Connect Claude Code once, then switch upstreams in the gateway; Hide API keys from a relay before requests leave your machine; Cut off dangerous tool calls from a relay before the client runs them · GitHub
Jev Harness
JevHarness is a Python project where an LLM writes a task-specific harness, meaning code and Jev judgment questions that turn task observations into actions, with optional reward-based reflection to improve it.
4 use cases: Draft a task-specific harness in Claude Code; Inspect the Pokémon harness archive locally; Install the Jev Harness skill in Codex · GitHub
Agent Memory
agent-memory is a Python runtime that gives AI agents long-term memory stored as plain Markdown files, with local ranked retrieval and a sleep-time consolidation layer, and Claude Code and Codex share one store.
5 use cases: Share one memory of decisions across Claude Code and Codex; Recover why a past decision was made and what it said then; Correct a stale memory without losing its history · GitHub
Neo Horse
NeoHorse is an open-weight model family from TokenRhythm for agent workflows, with 4B and 9B text models for tool use, coding, and instruction following, plus a 4B decision model called NeoHorse-Jev.
4 use cases: Serve NeoHorse-1 locally for agent tool calls; Route requests with NeoHorse-Jev structured decisions; Run a quantized NeoHorse-1 on your own hardware · GitHub
Eido
Eido is a macOS AI coding workspace, built on Zed and pi, for understanding code, working with an agent, and reviewing its changes.
3 use cases: Investigate a bug and review a proposed fix; Resume an interrupted coding task with its context; Ask the agent to inspect a web page beside your code · GitHub
DSH Agent Team Gui
A DeepSeek Harness Web profile plugin that saves reusable multi-model agent teams, runs planned workflows with per-member models and tools, and reports run status and Token usage.
4 use cases: Run a planner, engineer, and reviewer team on a small task; Save a team once and reuse it across projects; Check how many Tokens a multi-agent run used · GitHub
Chat on Steroids
Chat On Steroids is an independent, open-source local workspace that connects ChatGPT to your files, terminals, browser and desktop through MCP, with tools for delegation and long-task control.
3 use cases: Let ChatGPT edit and test a local project; Split independent jobs across parallel workers; Keep a long task on track across fresh chats · GitHub
Trueforge
TrueForge is an open-source TypeScript agent harness that runs the agent loop for you, covering model calls, MCP tools, skills, sandboxing, approvals, context management, and session state.
4 use cases: Run a local agent chat with your own models and tools; Script agent sessions with the TrueForge TypeScript SDK; Run a saved agent on a schedule and review each session · GitHub
Sentio
Sentio is a Rust email server that gives each AI agent its own email address, delivers inbound mail as structured webhooks, and sends replies over a REST API.
4 use cases: Give an agent its own email inbox and webhook feed; Host isolated email for each customer on one deployment; Let a coding agent send and read email through MCP tools · GitHub
Reslop
reslop is a terminal tool that lets you review AI-generated code and changes, leave findings and a backlog, and hand a repair plan to a local agent.
5 use cases: Review AI-written changes and run a repair plan in an agent; Review a GitHub PR or GitLab MR in a local terminal view; Import all GitHub issues into a local task backlog · GitHub
Simplify Codebase
An Agent Skill that helps coding agents find and remove accidental complexity in an existing codebase while keeping still-valid behavior, boundaries and compatibility intact.
4 use cases: Audit a whole repo for safe simplification candidates; Decide if readiness flags are duplicate state; Review and merge simplification findings from a PR · GitHub
AI Memory Skillpack
A bounded project-memory skill pack that gives Codex CLI, and optionally Claude Code, a size-capped, attributed memory of each repository, installed by an agent-executable installer.
5 use cases: Give Codex a durable, size-capped memory of your repo; Resume a repo in a new session from one handoff page; Check pack integrity and tests before trusting an install · GitHub
Cdaf
CDAF is an open sidecar format and toolkit that stores a timestamped text description of a video next to the file, so AI agents can read that text instead of re-analyzing the footage.
4 use cases: Stop coding agents from re-watching the same footage; Find b-roll clips in a library by plain-text search; Gate a build or pipeline on fresh video sidecars · GitHub
Neuroarxiv
NeuroArxiv is a TypeScript skill and CLI that checks real arXiv papers before Claude or Codex designs a new architecture, then returns one cited recommendation with a first step and known pitfalls.
3 use cases: Check arXiv prior art before designing a new architecture; Review a planned design in Codex against arXiv prior art; Get a cited algorithm brief from the terminal · GitHub
Quicksilver
Quicksilver is a Claude Code skill that hands bulk yes/no, label, and ranking judgments to Jev, a separate model, so Claude reads less and spends fewer tokens.
4 use cases: Shortlist the files that handle a concern before Claude reads them; Find the real failures in a long log without reading every line; Sort a batch of tickets or issues into labels in one pass · GitHub
Uebox
UE Box is an open-source TypeScript desktop AI agent that operates the Unreal Editor directly, working on levels, Blueprints, materials, and C++.
4 use cases: Let Codex or Claude Code edit an Unreal project via MCP; Ask an AI chat questions about an Unreal level; Fix Blueprint and C++ compile errors with a goal checklist · GitHub
Ufo Core
UFO is an open source Python runtime that lets a team run AI agents in chat, with the agents reading files and running commands, either on self-hosted infrastructure or through UFO's hosted service.
5 use cases: Run a shared team agent server on one machine; Hand a task to an agent in a hosted sandbox; Move a team UFO server onto Postgres · GitHub
Overtime
Overtime is a TypeScript app that runs long-lived coding agents from a brief, letting them work on their own across sessions and message you only when a decision is yours.
5 use cases: Keep CI and dependency PRs green while you work; Review each PR as it opens, worst findings first; Wake an on-call agent the moment logs or metrics spike · GitHub
Unreal Agent
Unreal Agent is a Go library and set of programs for building async-first AI agents, including a cloud agent runner and a terminal UI.
4 use cases: Run async agent tasks on the cloud runner; Run a coding agent from the terminal TUI; Build a custom agent tool on the harness library · GitHub
Rune
Rune is a GPU-accelerated IDE and terminal multiplexer for Go developers that can also run headless on a remote machine and serve workspaces to your other devices.
4 use cases: Open projects from a home machine on your laptop; Install the Rune IDE on macOS or Linux; Add a community extension from a Git repository · GitHub
Useagent
UseAgent is an open-source, self-hostable workspace that runs Claude Code, Codex, OpenCode or Pi in isolated cloud workspaces, where you hand off tasks from Slack, the web or the API and get files and pull requests back.
5 use cases: Hand off a code fix and get a pull request back; Self-host the agent workspace on your own Linux server; Run a research report or spreadsheet with a free model · GitHub
P3 Stack
p3-stack is a set of shell-installed agent skills that rebuild the pstack engineering workflow for T3 Code, using its delegation, worktree, PR watching, and scheduling features.
5 use cases: Run a rigorous coding task through the p3 mode router; Split a feature across child agents in separate worktrees; Get a thread woken when a PR checks, comments, or conflicts · GitHub
DSH Plugin Subscriptions
A plugin for the DeepSeek Harness (DSH) that lets you use ChatGPT (Codex), Claude, Grok, GitHub Copilot, and Google Antigravity subscriptions as LLM providers through OAuth login instead of API keys.
4 use cases: Run DSH sessions on a ChatGPT subscription; Generate and edit images with GPT or Grok; Check subscription usage before a long run · GitHub
Awesome Jev Typesafe
A curated list of tools, SDKs, guides, and articles for building with TypeSafe's Jev, a model that returns typed, calibrated answers instead of text.
5 use cases: Give a coding agent the Jev list as a skill; Make a first typed Jev decision from Python or TypeScript; Gate a cheap model's answers with a Jev check · GitHub
Poly Strata
PolyStrata runs large AI models on your own Windows or Linux PC and serves them through a localhost OpenAI or Anthropic compatible API, so nothing leaves the machine.
4 use cases: Run Claude Code on a local model with no cloud calls; Point existing AI scripts at a private local API; Ask questions about pictures without uploading them · GitHub
Eve Software Factory Template
Foreman is an open template for an AI software factory that takes tasks from GitHub and Linear, runs them through four AI agent stations, and delivers a reviewed draft pull request for a person to review and merge.
4 use cases: Turn a labeled GitHub issue into a reviewed draft PR; Run a factory task from your machine with the dev TUI; Delegate a Linear task to the factory and track it in Linear · GitHub
Skills
The skills CLI is a TypeScript command-line tool (run as npx skills) for finding, installing, updating, and creating reusable instructions that teach coding agents how to perform tasks.
5 use cases: Install a skill into a project for Claude Code; Try an agent skill without installing it; Keep installed agent Skills up to date · GitHub
Vicoa
Vicoa is an open-source orchestrator that runs a team of coding agents from desktop, mobile, and remote machines, and can be self-hosted.
4 use cases: Run several coding agents on one repo at once; Check and prompt running agents from your phone; Run coding agents on a VPS or remote server · GitHub
Nopus
nopus is a set of deterministic prose checks that asks coding agents for one clearer rewrite when a completed answer becomes abstract or hard to read.
4 use cases: Get clearer Claude Code answers via the Stop hook; Add clearer-response checks to Codex sessions; Filter abstract Pi responses with a rewrite request · GitHub
Vela Term
VelaTerm is a native desktop app that combines a terminal, session tree, and agent workspace for managing coding agent sessions, reachable from a browser and phone.
5 use cases: Run several coding agents side by side in one app; Split a large coding task across a team of agents; Let agent sessions search and message each other · GitHub
Official MCP Servers
A curated directory of official MCP servers from the companies behind products, with links to each server's repository or vendor docs.
4 use cases: Debug a failed CircleCI build from your AI assistant; Investigate a Grafana alert with an AI assistant; Query and branch Neon Postgres from an AI assistant · GitHub
Forward Implementation First
A Markdown agent skill, installed with a shell script, that makes coding agents ship real work first and skip self-invented bookkeeping like hashes, locks, and receipts.
4 use cases: Unblock a pipeline stage stalled by a stale marker; Keep an agent building features over multi-day roadmaps; Coordinate parallel agents with one writer for publication · GitHub
HERO Anti Over Defense
HERO is a short block of rules, pasted into a coding agent's config file, that stops the agent from over-defending: adding hashes, edge-case handling, review loops and extra scaffolding that the actual work does not need.
4 use cases: Stop Claude Code from over-hardening a small feature; Add the HERO block to Codex's AGENTS.md; Quote a HERO case ID when an agent defends its extra checks · GitHub
Architecture Canvas
Architecture Canvas is a local Python and Node app that lets you chat with your documents to create and refine editable architecture diagrams on an Excalidraw canvas.
3 use cases: Turn a requirements document into an editable architecture diagram; Edit and export a sample architecture diagram with no API key; Let an AI agent render diagrams through the MCP server · GitHub
Awesome Claude Mods
A community-maintained, evidence-graded index of Claude Code mods, plugins and hooks, rebuilt every two hours.
3 use cases: Find a Claude Code mod that matches your task; Turn on the built-in You should know mod; Review a community mod before you install it · GitHub
Awesome Jev Live
A curated, evidence-graded index of SDKs, MCP tools, agents, apps and open models built around TypeSafe AI's System One model, Jev.
4 use cases: Pick the right official Jev SDK for your stack; A/B test a typed Jev decision against a prompt; Shortlist Jev clients using evidence grades only · GitHub
Whirl
Whirl is an open source, MIT licensed AI chat app built on Next.js and Convex that routes many models through OpenRouter, with memory, artifacts, and MCP tool integrations.
4 use cases: Run your own copy of the Whirl chat app; Give a team chat access to its own tools via MCP; Keep sensitive chats locked on your device · GitHub
Halo
Halo is an open-source framework from White Circle for training large language and multimodal models with pre-training, SFT, preference optimization, and multi-turn RL on Hugging Face models.
5 use cases: Fine-tune a Hugging Face LLM with LoRA on one GPU; Train a mixture-of-experts model across 8 GPUs; Run multi-turn RL against vLLM or SGLang rollouts · GitHub
Oh My Cpa
Oh My CPA is a self-hosted web console that manages a CLIProxyAPI gateway's providers, credentials and client keys, and records the usage and cost of every request.
4 use cases: Install Oh My CPA by handing the setup to an agent; Track request volume, tokens and cost per model; Add providers and client keys behind one gateway · GitHub
Spark X2 6
Spark-X2.6-4B is a compact, general-purpose language model for on-device agents, released with Transformers, SGLang, vLLM, and Ollama deployment paths.
3 use cases: Run Spark-X2.6-4B locally in Python with Transformers; Serve Spark-X2.6-4B as an API with SGLang on NVIDIA GPUs; Serve Spark-X2.6-4B with vLLM for an OpenAI-style API · GitHub
DSH Config Manager
A DeepSeek Harness (DSH) plugin that backs up, restores, exports, imports, migrates and syncs a DSH configuration, including plugins, MCP servers, skills and workspaces.
5 use cases: Move your full DSH setup to a new computer in one export and import; Keep a rolling full DSH backup without manual exports; Keep DSH configuration in step between machines through Git, WebDAV, S3 or Gist · GitHub
Codex with ChatGPT
A bridge that lets Codex use the ChatGPT web app as its planning and review brain, while Codex keeps doing the code execution.
3 use cases: Plan a coding task in ChatGPT before Codex edits; Have ChatGPT review Codex changes without uploading the repo; Keep .env and key files out of ChatGPT reviews · GitHub
Openchatx MCP
OpenChatX is a local agent runtime that gives ChatGPT tools to edit files, run shell commands, control your computer, and use MCP servers through one official MCP connection.
4 use cases: Let ChatGPT edit code and run shell commands locally; Continue long ChatGPT tasks in a fresh conversation; Enforce coding rules and reusable workflows for ChatGPT · GitHub
AI Goodpractice
A hand-written guide to setting up an AI-driven coding workflow with rules, skills, MCPs and documentation, plus one MCP for tracking project progress.
3 use cases: Set coding style rules for a repo from existing code; Keep an AI on a project roadmap with project progress MCP; Write feature docs so AI finds code without full reads · GitHub
Baize
Baize is an open-source coding agent CLI, written in Python, that talks to several LLM backends, redacts sensitive data before it reaches the model, and can drive SPSS and SQL databases through MCP servers.
4 use cases: Refactor code across a whole repo from one plain-language request; Analyze SPSS .sav data files by describing the analysis; Query a SQL database while redacting personal data before the LLM sees it · GitHub
Voice Mem
VoiceMem is an open source memory system for real-time voice agents that stores facts and emotional context from speech and feeds relevant memories to a reply model.
5 use cases: Give a custom voice agent long-term memory; Store and search text facts with Voice Mem; Run the LoCoMo benchmark on a memory system · GitHub
Kyojin
Kyojin is an open source inference engine, built on ExLlamaV3, that runs large GLM, Qwen and MiMo models on one AMD Strix Halo mini PC with 128 GB of memory.
5 use cases: Run a local OpenAI-style LLM server on AMD; Read a 500K-token document in one prompt; Benchmark a Strix Halo machine and share results · GitHub
My Claude Code Mods
A set of custom Claude Code mods (plugins) that profile sessions, catch context waste and loops, record file history, and support verification and handoff work.
4 use cases: Find where a Claude Code session spends time and context; Find the edit that broke a test and restore the file; Catch agent retry loops before they burn your session · GitHub
Awesome Jev
A curated GitHub list of public projects, integrations, and discussions built on Jev, TypeSafe AI's System One model for typed decisions.
4 use cases: Find Jev projects that match your coding agent; Vet a listed Jev project before you adopt it; Spot same-day bulk submissions before trusting them · GitHub
LLM2 Jev
LLM2Jev is an open-source Python project that turns a local language model or multimodal model into a decision model that scores options, rubrics, yes/no questions, and image boxes using prefill only, with no training.
5 use cases: Route support tickets to a department with probabilities; Score drafts against a rubric and yes/no checks; Pick the next game action from an image · GitHub
Fireworks Open Eli5
An open Agent Skill for Codex and Claude Code that turns a question about a system into an evidence-backed, offline interactive HTML explainer.
3 use cases: Map a request path through a repo as an interactive story; Build an incident walkthrough with failure paths and fallbacks; Export an explainer as slides or a document for a briefing · GitHub
DSH Routing Suite
A Windows-oriented bundle of three DSH plugins: a runtime injector, task-aware reasoning-mode router presets, and a graded task protocol with a session measurement script.
4 use cases: Install the full routing suite with one command; Install the router presets by hand from a clone; Measure your own session with the graded metrics script · GitHub
Route VSCODE
A VSCode extension and local web dashboard that route GitHub Copilot Chat through a local proxy to the 9Router AI gateway, so the active model can be switched without reloading the window.
3 use cases: Switch Copilot Chat models without reloading VSCode; Compare AI model replies and latency side by side; Set up the Copilot model connector with one script · GitHub
Doc2 Harness
Doc2Harness is a Python benchmark that measures how well a model builds a working coding agent harness, then tests that harness against a scripted mock model using 94 released scenarios.
4 use cases: Score an existing coding agent harness; Re-run one scenario family to debug a harness; Confirm the scenario catalog is intact before scoring · GitHub
ZCode
ZCode is Z.ai's AI coding workbench, offering a desktop app, a browser interface and a terminal agent, all built from one TypeScript monorepo.
4 use cases: Run a browser coding workspace on a local project; Run the terminal agent from source on a project; Package the zcode command and test it locally · GitHub
Awesome Deep Seek Harness Plugins
A curated, daily-updated directory of public plugins, tools, clients, runtimes, and references for the DeepSeek Harness agent harness.
4 use cases: Run DSH on a local Ollama model with cloud fallback; Review DSH turns with a second model's advisory notes; Assign different models to DSH planner and subagent roles · GitHub
DeepSeek Harness Linux
A community fork that packages the DeepSeek Harness desktop app for Linux as AppImage, .deb, and .tar.gz files for x64 and arm64.
4 use cases: Install DeepSeek Harness desktop on Ubuntu from .deb; Run DeepSeek Harness from an AppImage that updates itself; Unpack DeepSeek Harness as a portable build on Raspberry Pi 5 · GitHub
Veri Ink
VeriInk is a Python tool that renders deterministic synthetic handwriting worksheets from model failures, recording seeds, labels, ink bounds, and file hashes for audit.
3 use cases: Rebuild the audit set and verify every hash; Turn a model failure into a reproducible handwriting batch; Regression-test an OCR model on six synthetic hard cases · GitHub
DSH Android
DSH Android is a DeepSeek Harness plugin that streams a live Android emulator or USB phone into a conversation and lets an agent build, run, and drive it through adb.
4 use cases: Build and run a Gradle app on a live Android device; Check a screen flow by element ID or on-screen text; Confirm a like or count changes by exactly one · GitHub
DSH iOS
DSH iOS Simulator is a DeepSeek Harness plugin that gives an agent a live iOS Simulator, and a USB-connected iPhone, inside the conversation, with 22 tools to boot, build, and drive the device.
4 use cases: Build an Xcode app in the simulator and check the result; Tap app controls by label and confirm the result; Check that a list row counter moves after a tap · GitHub
Other categories · All use cases · Hot repos
Questions
What is the best AI repo for Software and DevOps in 2026?
Right now the top pick on Deduped is Harness2: Harness² is a Python framework from Google Research that refines the harness around a fixed model (prompts, tools, memory, skills, orchestration) so it performs better on agent tasks.
How are these Software and DevOps repos ranked?
By how hot they are on GitHub (stars gained and mentions in the AI sources we track), then by how many practical use cases we found. The list is rebuilt every day.
Get 5 new use cases every morning. Free, one short email, unsubscribe anytime.
Get them free