Showing all 94
Runs Locally
94 records filed under this subject, newest first, each with what it is, what it cost and where the builder got stuck.
Records
94150Minecraft Crossover BridgePlay real Minecraft, camera and blocks and all, inside Monster Hunter: World or Elden Ring on an Apple Silicon Mac.149PhotoCraftAn image editor that aims at Photoshop feature by feature, written from scratch in Rust, with every action exposed as a command so the menus, a CLI, a JSON control channel and an agent all drive the same engine.ClaudeGitHub Copilot
148JeffA 0.8B open model that answers first and hands the query to a large model only when its own confidence falls below a threshold.Claude146LayaSmall encoder models that read a document once and return a probability for every option of every typed question, with nothing generated and a router that picks the checkpoint by script.ClaudeGitHub Copilot
145KevA family of small models that read one document and answer a typed set of questions with a probability for every option. No text is generated.Devin
144typesafe-computer-useIt drives a Mac toward a goal typed in plain English: OCR and the accessibility tree read the screen, a TypeSafe classifier picks the next action, and a writing model runs only for free text. A decision costs about $0.0002, a fiftieth of a cent, against $0.032 for a frontier model reading the same screenshot.ClaudeClaude CodeGPT+1
142DSCODEA macOS terminal coding agent built on DeepSeek Harness, carrying its own pinned harness profile, a terminal package it patches at build time, and a session layer: sessions that message each other, Ultra sub-agents, reviewed approvals, scheduled runs.DeepSeek HarnessDeepSeekClaude141ForbricA Minecraft mod loader that runs Fabric, Forge and NeoForge mods in the same instance.Claude
140Agents UniverseAn open-source agent platform that keeps one project context shared by everyone working in it. Agents read the whole knowledge base when a project is opened and write what they learn back into the same files while they work; knowledge is Markdown on disk with a database index behind it, and there is no embedding model or vector search.ClaudeClaude CodeGemini+1
138AI Second BrainA guide, a starter vault, 18 skills and 72 slash commands for a markdown knowledge base that files, links and lints itself while you keep reading.ClaudeClaude CodeGitHub Pages
137CodenotchA macOS agent app that turns a small black notch on a screen edge into a live tally of the coding assistants you have running: a ring per provider for how much of its limit is already spent, a hover card listing every limit window and its reset time, and a pulse inside the ring that says whether an agent is working or has stopped to ask you something.ClaudeClaude CodeCodex+1
136OKF Agent MemoryKeeps what a coding agent learns as plain Markdown inside the repository (an OKF v0.2 knowledge bundle searched in-process by BM25) so the memory can be diffed and reviewed instead of living in a database.ClaudeClaude CodeCodex+4
135AI EmployeesEight open source AI Employees, each a folder of scheduled routines that covers a whole business role, running on your own machine on the agent you already use: they drive the browser you are signed in to the way you do, draft and stage the work and leave the last click to you, get better at your business on every run, and every file they touch is yours.ClaudeClaude CodeCodex+3
134agent-memoryA long-term memory runtime for AI agents that keeps plain Markdown files as the single source of truth, ranks them locally without calling a model, answers recall with file paths the agent opens one level at a time, writes at conversation boundaries rather than on the agent’s initiative, and runs an independent sleep-time layer that may add and update on its own but can only ever file a deletion as a proposal — one store shared by Claude Code, Codex CLI and Hermes, with no API key.ClaudeClaude CodeCodex
133ReverifyA verifier for an AI’s claims about binaries: the model proposes, a deterministic toolkit checks the claim against the actual bytes, and what comes back is VERIFIED, REFUTED or INCONCLUSIVE with the evidence it read, and the verdicts, not the prose, are what survives a reset.ClaudeClaude CodeCodex+1
132PawWorkA Chrome extension that turns the page you are already logged into into the starting point of a task: you select the part of the live page that matters, describe the outcome you want in the side panel, and the agent works from that selection and hands back an editable office file, with your own model key, no hosted service, and the model-written JavaScript confined to a QuickJS sandbox inside the browser.Cursor
131EaselAn open-source content workbench for social media creators: one agent runs the whole loop — aggregate the hot lists, plan a topic, generate the copy, the cards, the voice and the video, publish the finished file to an account that is already logged in on seven Chinese platforms, then read the numbers back into the account profile that shaped the next round.ClaudeClaude CodeGemini+3
129Codex with ChatGPTA loopback bridge that hands planning and review to the ChatGPT web app while Codex keeps the keyboard: ChatGPT reads the workspace through nine read-only MCP tools, Codex types sub-kilobyte state messages into the ChatGPT tab, and no file body, diff or log is ever pasted into the chat.CodexGPTCursor+1
127LemmalogA Rust Datalog engine that treats agent memory as a deductive database, not a bigger vector store: facts asserted at the extraction boundary, stratified rules deriving closures and temporal views, provenance back to the source episode on every derived fact, and views maintained one epoch at a time — served to Claude Code and Kimi CLI as twelve MCP tools.ClaudeClaude CodeGPT
125RomeA self-hosted agent runtime built to grow the environment around a model: a Rome App packages an interface, executable actions, on-demand skills and a private database as git-tracked code an agent can reuse later, while the runtime gives every delegated subagent a child session of its own, forks a turn without letting it mutate its source, and fails a turn rather than quietly swapping the model that produced the conversation.ClaudeClaude CodeCodex+1
124Chat On SteroidsA desktop workspace that gives ChatGPT local tools over MCP (files, a shell, terminals, the desktop) plus a companion Chrome extension that drives the user’s own ChatGPT conversation through a tunnel, so the model works on a real project while the app records every tool call and moves the session into a fresh chat when the old one fills up.ClaudeClaude CodeCodex+1
120Tiny EngineerA 3D-printed desk robot that reads, thinks and types while your AI coding agent works — driven by a REST endpoint that any tool able to make an HTTP request can call.Claude CodeCursor
118WhiteboardA desktop app that puts an agent and a person on the same canvas: the agent draws the review (sequence diagrams, entity relationships, code peeks pinned to specific commits) and every shape on that canvas links back to the code it was drawn from.ClaudeClaude CodeCodex+4
117agy-staffA plugin that hires Google’s Antigravity CLI as a member of staff: the host agent (Claude Code, Codex or Pi) keeps the decisions and hands the surveys, reviews and scoped edits to a Gemini 3.8 Flash worker that runs in the background and answers through a job id.Google AntigravityClaudeClaude Code+2
116WakeA native desktop app that reads the private session directories of every coding agent on the machine, read-only, and gives them one window: browse by agent or project, full-text search that treats Chinese text and code substrings alike, one-click resume in the original project directory, a bundled MCP server and a command line so the agents themselves can read the history back, and remote hosts mirrored over SSH.ClaudeClaude CodeCodex
115Autoprompt SkillAn installable workflow for coding agents rather than a standalone program: it takes one goal, plus the constraints and the definition of done, and runs the execution loop (scope, plan, build, test, review, repair, verify) across eleven different agent tools. Coordination, execution and independent judgment are kept in separate layers, so the agent that writes a change is not the one that signs it off. Its headline claim, 45% fewer failures, comes from one Terminal-Bench 2.1 comparison the author ran, whose baseline verdict file is linked but not present in the repository.ClaudeClaude CodeCodex+2
114OpenBotAn open-source platform from CopilotKit that gives each AI coworker a computer of its own (a container with Chromium, a workspace volume and its own logins) where a coworker is any endpoint speaking AG-UI, and every browser, file, MCP or shell action passes through one gateway that decides it against a CEL policy, writes an audit row and only then acts, or refuses and names the rule.ClaudeClaude CodeCodex+1
113ProcoderA single Go binary that gives a coding agent the discipline of a senior engineer: a commit gate that treats a check it could not run as a failure, controllers that refuse to call unfinished work finished, and a loop that closes the class of every bug that escapes. It works with more than twenty agents, needs no runtime dependencies, and never edits a file behind the agent’s back.Claude CodeCodexCursor+5
112EverRoomA local-first Electron workspace where files, meetings, repositories and connected accounts are filed into a Context Room: every ingested source keeps an identity, a version and a provenance record, wiki pages and memory are derived from that ledger under a policy snapshot that is itself recorded, and an agent works only inside the Room it was handed.ClaudeClaude CodeCodex+1
111DeepSeek HarnessDeepSeek’s agent harness, built so that the model adapter, the tool registry, the session log and the agent loop itself are plugins, swapped from a configuration file rather than a fork.DeepSeekClaude CodeCodex
109VibeGameDescribe a game in one sentence and a team of agents divides the work — an architect plans it, a programmer builds it, an auditor checks the code against the plan, and a player has to actually play it before the task is accepted.Claude CodeCodex
108OpenMausBotAn open-source chat app where every bot in the sidebar is a real agent running through the CLI already installed on your machine, and each one can be handed a computer: a cloud Linux desktop, a Docker or Podman container on the same host, a container on a VPS you own, or the machine in front of you where the platform can show that it is safe.ClaudeClaude CodeCodex+4
107pgbotA single static Go binary that connects to PostgreSQL read-only, reads the server’s own statistics views and prints a graded, findings-first health report — and, because every run saves a local baseline, tells you what changed since last time; the same deterministic findings are served to AI agents over MCP, and the optional AI layer may only explain them.ClaudeClaude CodeCodex+3
106claudish-to-englishA Claude Code plugin that shows a plain rewrite of every assistant message underneath the original — produced by a local ollama model by default, or by the codex CLI, the Anthropic API or any OpenAI-compatible endpoint, with a /claudish command that switches it on and off and changes the display mode, the style, the output language and the model mid-session. It is display-only: the transcript and Claude’s own reasoning always keep the original text.Claude CodeClaudeOllama+2
104HarnessRouterThe self-hosted, Apache-2.0 edition of HarnessRouter: it puts sixteen existing agent CLIs (Codex, Claude Code, Hermes, DeepSeek Harness and twelve more) behind one OpenAI Responses-compatible API, with sessions, streaming, files, cancellation and structured failures, and it carries the Unified Harness Protocol it implements together with the conformance suite that measures it.ClaudeClaude CodeCodex+6
103AOCI-CODEA Go CLI and a local MCP server that keep one plain-text map of a repository (one line per managed file, saying what it is responsible for, what has to be read with it, what it exposes and what must not break) so a coding agent reads the system once instead of re-reading the code for every task.ClaudeClaude Code
101Phone HarnessA Python library and an agent skill that put a real iPhone or Android phone under a coding agent’s hands: the agent reads the screen through OCR or the accessibility tree, taps and types through HID events or adb, and then looks again to check what happened. Nothing is installed on the phone and there is no jailbreak — on a Mac the phone is an iPhone Mirroring window, elsewhere it is adb over USB or Wi-Fi.ClaudeClaude CodeCodex+2
100LodyA workspace where a team shares the coding agents it already runs: connect a machine, bring Claude Code, Codex, Kimi or any other agent that speaks the protocol, and dispatch work from desktop, phone, web or terminal while sessions delegate to each other and the code stays on the machine its owner connected.ClaudeClaude CodeCodex+3
096Skill RecorderA macOS-first Electron desktop app that records a real work session (screen video, app and window switches, browser URLs, clipboard previews, an optional session-owned terminal and spoken narration) then hands the timeline to the bundled GitHub Copilot CLI, which reconstructs one overall intent plus an ordered list of steps and turns the approved version into a SKILL.md procedure or a scheduled automation for Microsoft Scout, Microsoft 365 Copilot Cowork or Copilot Studio.GitHub CopilotWhisper
095ripwireA C++23 command-line tool and MCP server that parses a repository into a ranked call graph, then answers what a symbol reaches, what a change breaks, which tests reach it and what the change made worse.ClaudeClaude CodeCodex
094agentacctReads the session files your coding agents already write, records what they claim they did over MCP and hooks, prices the tokens from a public price list, and shows the whole thing in a local dashboard that never leaves the machine — keeping an agent’s word for done and a passing check as two separate facts.ClaudeClaude CodeCodex+2
093TrueForgeA self-hosted runtime that takes over the agent loop (model calls, MCP tool servers, git-backed skills, an optional sandbox, human approvals, context compaction and session state) and hands it back three ways: a chat UI, an HTTP API with a TypeScript SDK, and an embeddable UI SDK, on SQLite in a single process or on Postgres and Redis when it is hosted.Claude CodeCursorRailway+1
091ZeronA Rust desktop app that runs the coding agents you already use (Claude Code, Codex, Cursor, Devin, Grok, Hermes, Pi and Antigravity) on your own machine, with no account required, and syncs the sessions to your other devices only if you sign in.Claude CodeClaudeCodex+5
089NovelToGameSeven skills that install into a coding agent and turn a whole novel into a game: they write a source bible in which every fact carries its chapter, kill candidate directions with named hard vetoes, lock a target runtime in a brief and build for it, and prove the build plays in one recorded run of six checks.ClaudeClaude CodeCodex+2
088MemmyA local memory hub: one service on the machine holds the traces, policies, world model and skills that Claude Code, Codex, Cursor, DeepSeek Harness, OpenClaw, Hermes and OpenCode all read and write, so context carries between agents instead of staying inside whichever one was open.ClaudeClaude CodeCodex+2
087Drama SkillsEleven independently installable agent skills that take an AI short drama from a source novel to episode scripts, a storyboard, image and video prompts, confirmed generation, a rendered cut and a review, with five Markdown documents per episode as the only creator-facing record.ClaudeClaude CodeCodex+2
086agent-managerTen coding agents, one terminal list: live status read from each tmux pane, a quick prompt that answers them without attaching, a dead session revived on its own conversation, and a diff review whose comments go back to the agent as one numbered round.ClaudeClaude CodeCodex+2
085OpenChatCutA local-first video editor whose editing surface is a conversation: the built-in agent and external Codex or Claude Code sessions call the same editing tools the interface itself uses, so every change lands on a real multi-track timeline as a clip, transition, caption, effect or audio item that can still be dragged, undone and exported. Projects and media stay on the machine, and preview and final render both come out of Remotion.Claude CodeCodexClaude+8
084Grok BuildGrok Build is SpaceXAI’s terminal coding agent — a full-screen, mouse-driven TUI over a harness that reads a codebase, edits files, runs shell commands, searches the web and keeps long-running tasks alive; the same binary also runs headless for scripts and CI and speaks the Agent Client Protocol to editors, it is extensible through hooks, plugins, skills and MCP servers, and it deliberately adopts the configuration of the agents it competes with while accepting no outside contributions at all.Claude CodeCodexCursor+1
083mindwalkA local Go binary that reads Claude Code, Codex and pi session logs and replays them as light moving across a deterministic 3D map of the repository, so the files an agent searched, read and edited glow and everything else stays dark.ClaudeClaude CodeCodex
080EngramA learning engine that installs into a coding agent: a curriculum architect breaks a topic into a first-principles concept map, a tutor makes you predict, attempt and explain before it explains, a blind assessor grades your verbatim free recall and writes a receipt for every verdict, and a deterministic FSRS-4.5 core in one Python file decides when each concept comes back, with explorable HTML built only for the concepts whose content rewards manipulation.ClaudeClaude CodeCodex+2
079GraftIt reads a repository with tree-sitter, writes what it finds as a folder of linked markdown nodes plus a per-symbol code graph, and hands the matching slices to whichever coding agent you already use: up front in the prompt, or through six MCP tools the agent calls itself.ClaudeClaude CodeCursor+1
078ColibrìAn inference engine in pure C with no engine dependencies that treats storage, RAM and VRAM as one hierarchy, so 744B to 2.8T-parameter mixture-of-experts models run on hardware people already own.ClaudeCodex
076delegate-skillsA skills package in which every coding-agent CLI gets its own delegation skill: the orchestrating agent writes a self-contained brief, a separate CLI edits a real working tree, and the human keeps the review and the commit.ClaudeClaude CodeCodex+6
075TuttiA desktop workspace where several coding agents share one set of files, tasks and app outputs, so context moves between Claude Code and Codex instead of through you.ClaudeClaude CodeCodex+2
073ReticleAn MCP server and a dev-only SDK that let a coding agent read and drive a running web or desktop app from the inside, then answer with a verdict and the file and line to fix instead of a screenshot.ClaudeClaude CodeCline+7
072makerskillsTwenty-one agent skills for running a one-person business (decide, unstuck, maker-council, deep-research, second-brain, company-brain, domain, jab-hook, pm, personal-cfo and the rest) each one a Markdown workflow document rather than a program, installed into Claude Code, Codex, Cursor or any host that reads the Agent Skill format, with every piece of personal state kept in a config directory the repository never touches.ClaudeClaude CodeCodex+3
071GSD CoreGit. Ship. Done. — a meta-prompting, context-engineering and spec-driven development framework that runs the same five-step loop on every milestone: discuss, plan, execute, verify and ship. The heavy work is pushed into fresh-context subagents so the main session stays lean, and every decision is written into Markdown and JSON under a planning directory instead of living in the conversation.Claude CodeClaudeCodex+7
069classic-vibe-macWrite C in a browser tab, press Build & Run, and a System 7 Macintosh boots with your freshly compiled 68k application sitting on its desktop.Claude Code
068book-to-skillA converter that reads a document (one file, a folder, a glob or a list of paths, in PDF, EPUB, DOCX, HTML, RTF, MOBI or plain text) and writes an agent skill: a core file of mental models with a chapter index, one file per chapter that loads only when a question touches it, and a glossary, a patterns file and a cheatsheet beside them.ClaudeClaude CodeCursor+2
067VibeRaveA fork of Strudel that lets you hold a key, say a genre out loud, and have the music change on the next bar without the beat ever stopping.Claude066REAA local MCP server and command line that turn decompilers, a JavaScript and .NET reader, a browser and a process recorder into one set of tools an agent can call, and return every finding with the evidence and the limits attached to it.ClaudeCursorCodex+1
065FridayA personal assistant that runs around the clock using Claude Code itself as the runtime: Telegram for conversation, MCP plugins for mail, notes, voice and git, and a cron system the assistant reorganises on its own.ClaudeClaude Code
063gate4agentA Rust workbench for running, watching and orchestrating CLI coding agents across machines. It wraps Claude Code, Codex, Kimi and Grok behind one transport, one relay and one operator protocol.ClaudeClaude Code
062codesightA command-line tool that reads a repository once and writes it back out as a handful of small markdown files (routes, schema, components, env vars, dependency graph) so a coding agent can start from a map instead of exploring the code itself.ClaudeClaude CodeCline+4059Blatant-WhyA protein-design campaign kit for Claude Code: prompt-defined agents write the model inputs, submit them to BoltzGen, Protenix or PXDesign on whatever GPU you have, and rank what comes back with a published interface score.Claude CodeClaude
058LeanCTXA local layer that sits beside a coding agent and decides what reaches the model: file reads are compressed and cached, command output is compressed by per-command rules, session findings persist across chats, and a local proxy rewrites each request without breaking the provider’s prompt cache — with a savings ledger, a budget and a dashboard for what it measured.ClaudeClaude CodeCursor+2
057AI Job SearchA job-application framework that runs on your own machine: it scores postings against a profile you fill in, drafts a tailored CV and cover letter in LaTeX, compiles them, reads the rendered PDF back, and stops one step short of sending anything.ClaudeClaude Code
055OpenWolfA local .wolf folder that holds one project’s task state, file index, notes and known fixes for whichever coding agent is running, injects only the changed parts of it at session start, on a prompt and around compaction, and reports the provider token counters it reads back out of Claude transcripts, Codex rollouts and OpenCode plugin records.ClaudeClaude CodeCodex+4
052AI Comic BuilderA self-hosted pipeline that takes a script to a finished animated episode: parse it, extract the characters, generate four-view reference sheets, storyboard it, generate keyframes per shot, then interpolate the video and burn in subtitles.Claude
051ORCHA runtime for running several coding agents on one project at once: you define a team, give it a goal, and a CTO agent decomposes the work while the rest pick up tasks (across Claude, Codex, Cursor, Grok and a plain shell) with all the state kept in files rather than a database.ClaudeClaude Code
049vphone-cliRuns iOS inside Apple’s Virtualization.framework on a physical Apple Silicon Mac, with a graphical window, pre-patched firmware, VM cloning and an optional local HTTP and WebSocket automation API.Claude
047N.O.R.A.CoreA single-author AI companion with two brains and an encrypted diary, whose earliest commits are signed in the assistant’s own name.Claude Code
046Neko MasterA self-hosted dashboard that shows where a home gateway’s traffic actually goes (by domain, by IP, by proxy node) reading it live from Clash, mihomo or OpenClash over WebSocket.ClaudeClaude Code
045Fud AIA calorie and workout tracker for iOS and Android that runs on your own AI key: photograph a meal, say it out loud or type it, and the analysis is done by whichever of thirteen providers you picked — with an on-device model as the last resort when all of them fail.CursorClaudeGitHub Copilot
044AIYOUA canvas of twelve nodes that carries an idea to a finished short drama: outline, episode split, character sheets, storyboards, keyframes, then video. A provider layer keeps the run alive by falling back down a priority list when a model’s quota runs out.ClaudeGeminiDeepSeek
043cocoindex-codeA command-line tool and MCP server that keeps a local semantic index of a codebase (tree-sitter chunks, embedding vectors in SQLite) so a coding agent can find code by description instead of reading files, with a background daemon holding the model and an index that only re-embeds what changed.ClaudeClaude CodeCodex+3
042Vibe ProjectsOne person’s monorepo of six AI-generated projects (an Android agent harness, a per-hunk diff reviewer for VS Code, and four small Android apps) kept deliberately as a running measure of what the models could build.ClaudeGPTClaude Code041JianYing Editor SkillAn agent skill that edits in JianYing by writing the app’s own project files, and drives the app for the parts that are not files.Google AntigravityClaude CodeCursor
040CodewhaleA terminal coding agent in Rust that reads your project, edits files and runs commands: a runtime shared by a web client and a desktop app, a model layer that treats providers as interchangeable, and a contributing guide that spends its length on how several agents share one checkout.ClaudeCursorGitHub Copilot
039CodeGraphA code knowledge graph for agents: tree-sitter parses the source, FalkorDB stores the structure and its history, and five MCP tools answer questions about it from a Claude Code session or a browser dashboard.ClaudeClaude Codev0
037HeadroomA compression layer that sits between an agent and the model it calls: tool output, logs, file reads, RAG chunks and conversation history are shrunk on the machine that runs it, and the originals are kept in a local store so the model can ask for any block back.ClaudeClaude CodeCline+7
035BrawloreA Diablo II homage in plain JavaScript (endless dungeons, ten equipment sets and a two-level branching skill tree) built without its author reading the code, and shipped under a name of its own.GeminiClaude Code
034MiroFishA prediction engine that builds a parallel world of hundreds of AI agents out of a document you upload, then runs it forward to see what happens next.ClaudeGemini
028CCManagerA terminal menu that keeps several coding agents running at once, one per git worktree: it shows which are busy, which are waiting on you and which are idle, creates and merges the worktrees, and can bring the whole set back after a crash.ClaudeClaude Code
024VibeFlow-P5A p5.js particle sketch, a 356-byte Go file server and a hardened Docker image, published alongside a manifesto for vibe coding, with the conversation’s German left in the comments.Claude
023gooseA general-purpose AI agent that runs on your own machine, shipped as a desktop app, a CLI and an API, with extensions built on the Model Context Protocol.ClaudeGPTOllama
020Moonshine VoiceA speech stack for voice agents and interfaces: transcription, intent recognition and text to speech, built to run at low latency on small devices, with 74 of its recent commits co-authored by Cursor.Claude
017BettaFishA public-opinion analysis system in which several kinds of research agent are made to argue with each other on purpose, so the output is not one model’s opinion written up at length.
011Open WebUIA self-hosted chat interface that talks to Ollama and to any OpenAI-compatible API.Ollama
010OllamaRuns open-weight models on your own machine and serves them on a local API.llama.cpp
009Open InterpreterA terminal coding agent that started as a local alternative to OpenAI Code Interpreter and was rebuilt in 2026 as a Rust fork of Codex, aimed at open-weight models.CodexClaude CodeDeepSeek
003ComfyUIRuns diffusion models from a node graph: connect loading, prompting, sampling, decoding and saving, then run the graph.Stable Diffusion
001whisper.cppOpenAI’s Whisper speech recognition rewritten in C/C++, so it runs on a CPU, a phone, or in a browser tab.Whisper
148JeffA 0.8B open model that answers first and hands the query to a large model only when its own confidence falls below a threshold.Claude146LayaSmall encoder models that read a document once and return a probability for every option of every typed question, with nothing generated and a router that picks the checkpoint by script.ClaudeGitHub Copilot
145KevA family of small models that read one document and answer a typed set of questions with a probability for every option. No text is generated.Devin
144typesafe-computer-useIt drives a Mac toward a goal typed in plain English: OCR and the accessibility tree read the screen, a TypeSafe classifier picks the next action, and a writing model runs only for free text. A decision costs about $0.0002, a fiftieth of a cent, against $0.032 for a frontier model reading the same screenshot.ClaudeClaude CodeGPT+1
142DSCODEA macOS terminal coding agent built on DeepSeek Harness, carrying its own pinned harness profile, a terminal package it patches at build time, and a session layer: sessions that message each other, Ultra sub-agents, reviewed approvals, scheduled runs.DeepSeek HarnessDeepSeekClaude141ForbricA Minecraft mod loader that runs Fabric, Forge and NeoForge mods in the same instance.Claude
138AI Second BrainA guide, a starter vault, 18 skills and 72 slash commands for a markdown knowledge base that files, links and lints itself while you keep reading.ClaudeClaude CodeGitHub Pages
137CodenotchA macOS agent app that turns a small black notch on a screen edge into a live tally of the coding assistants you have running: a ring per provider for how much of its limit is already spent, a hover card listing every limit window and its reset time, and a pulse inside the ring that says whether an agent is working or has stopped to ask you something.ClaudeClaude CodeCodex+1
136OKF Agent MemoryKeeps what a coding agent learns as plain Markdown inside the repository (an OKF v0.2 knowledge bundle searched in-process by BM25) so the memory can be diffed and reviewed instead of living in a database.ClaudeClaude CodeCodex+4
135AI EmployeesEight open source AI Employees, each a folder of scheduled routines that covers a whole business role, running on your own machine on the agent you already use: they drive the browser you are signed in to the way you do, draft and stage the work and leave the last click to you, get better at your business on every run, and every file they touch is yours.ClaudeClaude CodeCodex+3
134agent-memoryA long-term memory runtime for AI agents that keeps plain Markdown files as the single source of truth, ranks them locally without calling a model, answers recall with file paths the agent opens one level at a time, writes at conversation boundaries rather than on the agent’s initiative, and runs an independent sleep-time layer that may add and update on its own but can only ever file a deletion as a proposal — one store shared by Claude Code, Codex CLI and Hermes, with no API key.ClaudeClaude CodeCodex
133ReverifyA verifier for an AI’s claims about binaries: the model proposes, a deterministic toolkit checks the claim against the actual bytes, and what comes back is VERIFIED, REFUTED or INCONCLUSIVE with the evidence it read, and the verdicts, not the prose, are what survives a reset.ClaudeClaude CodeCodex+1
132PawWorkA Chrome extension that turns the page you are already logged into into the starting point of a task: you select the part of the live page that matters, describe the outcome you want in the side panel, and the agent works from that selection and hands back an editable office file, with your own model key, no hosted service, and the model-written JavaScript confined to a QuickJS sandbox inside the browser.Cursor
131EaselAn open-source content workbench for social media creators: one agent runs the whole loop — aggregate the hot lists, plan a topic, generate the copy, the cards, the voice and the video, publish the finished file to an account that is already logged in on seven Chinese platforms, then read the numbers back into the account profile that shaped the next round.ClaudeClaude CodeGemini+3
129Codex with ChatGPTA loopback bridge that hands planning and review to the ChatGPT web app while Codex keeps the keyboard: ChatGPT reads the workspace through nine read-only MCP tools, Codex types sub-kilobyte state messages into the ChatGPT tab, and no file body, diff or log is ever pasted into the chat.CodexGPTCursor+1
127LemmalogA Rust Datalog engine that treats agent memory as a deductive database, not a bigger vector store: facts asserted at the extraction boundary, stratified rules deriving closures and temporal views, provenance back to the source episode on every derived fact, and views maintained one epoch at a time — served to Claude Code and Kimi CLI as twelve MCP tools.ClaudeClaude CodeGPT
125RomeA self-hosted agent runtime built to grow the environment around a model: a Rome App packages an interface, executable actions, on-demand skills and a private database as git-tracked code an agent can reuse later, while the runtime gives every delegated subagent a child session of its own, forks a turn without letting it mutate its source, and fails a turn rather than quietly swapping the model that produced the conversation.ClaudeClaude CodeCodex+1
124Chat On SteroidsA desktop workspace that gives ChatGPT local tools over MCP (files, a shell, terminals, the desktop) plus a companion Chrome extension that drives the user’s own ChatGPT conversation through a tunnel, so the model works on a real project while the app records every tool call and moves the session into a fresh chat when the old one fills up.ClaudeClaude CodeCodex+1
120Tiny EngineerA 3D-printed desk robot that reads, thinks and types while your AI coding agent works — driven by a REST endpoint that any tool able to make an HTTP request can call.Claude CodeCursor
117agy-staffA plugin that hires Google’s Antigravity CLI as a member of staff: the host agent (Claude Code, Codex or Pi) keeps the decisions and hands the surveys, reviews and scoped edits to a Gemini 3.8 Flash worker that runs in the background and answers through a job id.Google AntigravityClaudeClaude Code+2
116WakeA native desktop app that reads the private session directories of every coding agent on the machine, read-only, and gives them one window: browse by agent or project, full-text search that treats Chinese text and code substrings alike, one-click resume in the original project directory, a bundled MCP server and a command line so the agents themselves can read the history back, and remote hosts mirrored over SSH.ClaudeClaude CodeCodex
115Autoprompt SkillAn installable workflow for coding agents rather than a standalone program: it takes one goal, plus the constraints and the definition of done, and runs the execution loop (scope, plan, build, test, review, repair, verify) across eleven different agent tools. Coordination, execution and independent judgment are kept in separate layers, so the agent that writes a change is not the one that signs it off. Its headline claim, 45% fewer failures, comes from one Terminal-Bench 2.1 comparison the author ran, whose baseline verdict file is linked but not present in the repository.ClaudeClaude CodeCodex+2
114OpenBotAn open-source platform from CopilotKit that gives each AI coworker a computer of its own (a container with Chromium, a workspace volume and its own logins) where a coworker is any endpoint speaking AG-UI, and every browser, file, MCP or shell action passes through one gateway that decides it against a CEL policy, writes an audit row and only then acts, or refuses and names the rule.ClaudeClaude CodeCodex+1
113ProcoderA single Go binary that gives a coding agent the discipline of a senior engineer: a commit gate that treats a check it could not run as a failure, controllers that refuse to call unfinished work finished, and a loop that closes the class of every bug that escapes. It works with more than twenty agents, needs no runtime dependencies, and never edits a file behind the agent’s back.Claude CodeCodexCursor+5
107pgbotA single static Go binary that connects to PostgreSQL read-only, reads the server’s own statistics views and prints a graded, findings-first health report — and, because every run saves a local baseline, tells you what changed since last time; the same deterministic findings are served to AI agents over MCP, and the optional AI layer may only explain them.ClaudeClaude CodeCodex+3
106claudish-to-englishA Claude Code plugin that shows a plain rewrite of every assistant message underneath the original — produced by a local ollama model by default, or by the codex CLI, the Anthropic API or any OpenAI-compatible endpoint, with a /claudish command that switches it on and off and changes the display mode, the style, the output language and the model mid-session. It is display-only: the transcript and Claude’s own reasoning always keep the original text.Claude CodeClaudeOllama+2
104HarnessRouterThe self-hosted, Apache-2.0 edition of HarnessRouter: it puts sixteen existing agent CLIs (Codex, Claude Code, Hermes, DeepSeek Harness and twelve more) behind one OpenAI Responses-compatible API, with sessions, streaming, files, cancellation and structured failures, and it carries the Unified Harness Protocol it implements together with the conformance suite that measures it.ClaudeClaude CodeCodex+6
103AOCI-CODEA Go CLI and a local MCP server that keep one plain-text map of a repository (one line per managed file, saying what it is responsible for, what has to be read with it, what it exposes and what must not break) so a coding agent reads the system once instead of re-reading the code for every task.ClaudeClaude Code
101Phone HarnessA Python library and an agent skill that put a real iPhone or Android phone under a coding agent’s hands: the agent reads the screen through OCR or the accessibility tree, taps and types through HID events or adb, and then looks again to check what happened. Nothing is installed on the phone and there is no jailbreak — on a Mac the phone is an iPhone Mirroring window, elsewhere it is adb over USB or Wi-Fi.ClaudeClaude CodeCodex+2
100LodyA workspace where a team shares the coding agents it already runs: connect a machine, bring Claude Code, Codex, Kimi or any other agent that speaks the protocol, and dispatch work from desktop, phone, web or terminal while sessions delegate to each other and the code stays on the machine its owner connected.ClaudeClaude CodeCodex+3
096Skill RecorderA macOS-first Electron desktop app that records a real work session (screen video, app and window switches, browser URLs, clipboard previews, an optional session-owned terminal and spoken narration) then hands the timeline to the bundled GitHub Copilot CLI, which reconstructs one overall intent plus an ordered list of steps and turns the approved version into a SKILL.md procedure or a scheduled automation for Microsoft Scout, Microsoft 365 Copilot Cowork or Copilot Studio.GitHub CopilotWhisper
095ripwireA C++23 command-line tool and MCP server that parses a repository into a ranked call graph, then answers what a symbol reaches, what a change breaks, which tests reach it and what the change made worse.ClaudeClaude CodeCodex
094agentacctReads the session files your coding agents already write, records what they claim they did over MCP and hooks, prices the tokens from a public price list, and shows the whole thing in a local dashboard that never leaves the machine — keeping an agent’s word for done and a passing check as two separate facts.ClaudeClaude CodeCodex+2
093TrueForgeA self-hosted runtime that takes over the agent loop (model calls, MCP tool servers, git-backed skills, an optional sandbox, human approvals, context compaction and session state) and hands it back three ways: a chat UI, an HTTP API with a TypeScript SDK, and an embeddable UI SDK, on SQLite in a single process or on Postgres and Redis when it is hosted.Claude CodeCursorRailway+1
091ZeronA Rust desktop app that runs the coding agents you already use (Claude Code, Codex, Cursor, Devin, Grok, Hermes, Pi and Antigravity) on your own machine, with no account required, and syncs the sessions to your other devices only if you sign in.Claude CodeClaudeCodex+5
088MemmyA local memory hub: one service on the machine holds the traces, policies, world model and skills that Claude Code, Codex, Cursor, DeepSeek Harness, OpenClaw, Hermes and OpenCode all read and write, so context carries between agents instead of staying inside whichever one was open.ClaudeClaude CodeCodex+2
086agent-managerTen coding agents, one terminal list: live status read from each tmux pane, a quick prompt that answers them without attaching, a dead session revived on its own conversation, and a diff review whose comments go back to the agent as one numbered round.ClaudeClaude CodeCodex+2
085OpenChatCutA local-first video editor whose editing surface is a conversation: the built-in agent and external Codex or Claude Code sessions call the same editing tools the interface itself uses, so every change lands on a real multi-track timeline as a clip, transition, caption, effect or audio item that can still be dragged, undone and exported. Projects and media stay on the machine, and preview and final render both come out of Remotion.Claude CodeCodexClaude+8
084Grok BuildGrok Build is SpaceXAI’s terminal coding agent — a full-screen, mouse-driven TUI over a harness that reads a codebase, edits files, runs shell commands, searches the web and keeps long-running tasks alive; the same binary also runs headless for scripts and CI and speaks the Agent Client Protocol to editors, it is extensible through hooks, plugins, skills and MCP servers, and it deliberately adopts the configuration of the agents it competes with while accepting no outside contributions at all.Claude CodeCodexCursor+1
083mindwalkA local Go binary that reads Claude Code, Codex and pi session logs and replays them as light moving across a deterministic 3D map of the repository, so the files an agent searched, read and edited glow and everything else stays dark.ClaudeClaude CodeCodex
080EngramA learning engine that installs into a coding agent: a curriculum architect breaks a topic into a first-principles concept map, a tutor makes you predict, attempt and explain before it explains, a blind assessor grades your verbatim free recall and writes a receipt for every verdict, and a deterministic FSRS-4.5 core in one Python file decides when each concept comes back, with explorable HTML built only for the concepts whose content rewards manipulation.ClaudeClaude CodeCodex+2
079GraftIt reads a repository with tree-sitter, writes what it finds as a folder of linked markdown nodes plus a per-symbol code graph, and hands the matching slices to whichever coding agent you already use: up front in the prompt, or through six MCP tools the agent calls itself.ClaudeClaude CodeCursor+1
078ColibrìAn inference engine in pure C with no engine dependencies that treats storage, RAM and VRAM as one hierarchy, so 744B to 2.8T-parameter mixture-of-experts models run on hardware people already own.ClaudeCodex
076delegate-skillsA skills package in which every coding-agent CLI gets its own delegation skill: the orchestrating agent writes a self-contained brief, a separate CLI edits a real working tree, and the human keeps the review and the commit.ClaudeClaude CodeCodex+6
075TuttiA desktop workspace where several coding agents share one set of files, tasks and app outputs, so context moves between Claude Code and Codex instead of through you.ClaudeClaude CodeCodex+2
073ReticleAn MCP server and a dev-only SDK that let a coding agent read and drive a running web or desktop app from the inside, then answer with a verdict and the file and line to fix instead of a screenshot.ClaudeClaude CodeCline+7
071GSD CoreGit. Ship. Done. — a meta-prompting, context-engineering and spec-driven development framework that runs the same five-step loop on every milestone: discuss, plan, execute, verify and ship. The heavy work is pushed into fresh-context subagents so the main session stays lean, and every decision is written into Markdown and JSON under a planning directory instead of living in the conversation.Claude CodeClaudeCodex+7
069classic-vibe-macWrite C in a browser tab, press Build & Run, and a System 7 Macintosh boots with your freshly compiled 68k application sitting on its desktop.Claude Code
068book-to-skillA converter that reads a document (one file, a folder, a glob or a list of paths, in PDF, EPUB, DOCX, HTML, RTF, MOBI or plain text) and writes an agent skill: a core file of mental models with a chapter index, one file per chapter that loads only when a question touches it, and a glossary, a patterns file and a cheatsheet beside them.ClaudeClaude CodeCursor+2
067VibeRaveA fork of Strudel that lets you hold a key, say a genre out loud, and have the music change on the next bar without the beat ever stopping.Claude066REAA local MCP server and command line that turn decompilers, a JavaScript and .NET reader, a browser and a process recorder into one set of tools an agent can call, and return every finding with the evidence and the limits attached to it.ClaudeCursorCodex+1
063gate4agentA Rust workbench for running, watching and orchestrating CLI coding agents across machines. It wraps Claude Code, Codex, Kimi and Grok behind one transport, one relay and one operator protocol.ClaudeClaude Code
062codesightA command-line tool that reads a repository once and writes it back out as a handful of small markdown files (routes, schema, components, env vars, dependency graph) so a coding agent can start from a map instead of exploring the code itself.ClaudeClaude CodeCline+4059Blatant-WhyA protein-design campaign kit for Claude Code: prompt-defined agents write the model inputs, submit them to BoltzGen, Protenix or PXDesign on whatever GPU you have, and rank what comes back with a published interface score.Claude CodeClaude
057AI Job SearchA job-application framework that runs on your own machine: it scores postings against a profile you fill in, drafts a tailored CV and cover letter in LaTeX, compiles them, reads the rendered PDF back, and stops one step short of sending anything.ClaudeClaude Code
055OpenWolfA local .wolf folder that holds one project’s task state, file index, notes and known fixes for whichever coding agent is running, injects only the changed parts of it at session start, on a prompt and around compaction, and reports the provider token counters it reads back out of Claude transcripts, Codex rollouts and OpenCode plugin records.ClaudeClaude CodeCodex+4
051ORCHA runtime for running several coding agents on one project at once: you define a team, give it a goal, and a CTO agent decomposes the work while the rest pick up tasks (across Claude, Codex, Cursor, Grok and a plain shell) with all the state kept in files rather than a database.ClaudeClaude Code
049vphone-cliRuns iOS inside Apple’s Virtualization.framework on a physical Apple Silicon Mac, with a graphical window, pre-patched firmware, VM cloning and an optional local HTTP and WebSocket automation API.Claude
047N.O.R.A.CoreA single-author AI companion with two brains and an encrypted diary, whose earliest commits are signed in the assistant’s own name.Claude Code
046Neko MasterA self-hosted dashboard that shows where a home gateway’s traffic actually goes (by domain, by IP, by proxy node) reading it live from Clash, mihomo or OpenClash over WebSocket.ClaudeClaude Code
045Fud AIA calorie and workout tracker for iOS and Android that runs on your own AI key: photograph a meal, say it out loud or type it, and the analysis is done by whichever of thirteen providers you picked — with an on-device model as the last resort when all of them fail.CursorClaudeGitHub Copilot
044AIYOUA canvas of twelve nodes that carries an idea to a finished short drama: outline, episode split, character sheets, storyboards, keyframes, then video. A provider layer keeps the run alive by falling back down a priority list when a model’s quota runs out.ClaudeGeminiDeepSeek
043cocoindex-codeA command-line tool and MCP server that keeps a local semantic index of a codebase (tree-sitter chunks, embedding vectors in SQLite) so a coding agent can find code by description instead of reading files, with a background daemon holding the model and an index that only re-embeds what changed.ClaudeClaude CodeCodex+3
042Vibe ProjectsOne person’s monorepo of six AI-generated projects (an Android agent harness, a per-hunk diff reviewer for VS Code, and four small Android apps) kept deliberately as a running measure of what the models could build.ClaudeGPTClaude Code041JianYing Editor SkillAn agent skill that edits in JianYing by writing the app’s own project files, and drives the app for the parts that are not files.Google AntigravityClaude CodeCursor
040CodewhaleA terminal coding agent in Rust that reads your project, edits files and runs commands: a runtime shared by a web client and a desktop app, a model layer that treats providers as interchangeable, and a contributing guide that spends its length on how several agents share one checkout.ClaudeCursorGitHub Copilot
039CodeGraphA code knowledge graph for agents: tree-sitter parses the source, FalkorDB stores the structure and its history, and five MCP tools answer questions about it from a Claude Code session or a browser dashboard.ClaudeClaude Codev0
037HeadroomA compression layer that sits between an agent and the model it calls: tool output, logs, file reads, RAG chunks and conversation history are shrunk on the machine that runs it, and the originals are kept in a local store so the model can ask for any block back.ClaudeClaude CodeCline+7
035BrawloreA Diablo II homage in plain JavaScript (endless dungeons, ten equipment sets and a two-level branching skill tree) built without its author reading the code, and shipped under a name of its own.GeminiClaude Code
034MiroFishA prediction engine that builds a parallel world of hundreds of AI agents out of a document you upload, then runs it forward to see what happens next.ClaudeGemini
028CCManagerA terminal menu that keeps several coding agents running at once, one per git worktree: it shows which are busy, which are waiting on you and which are idle, creates and merges the worktrees, and can bring the whole set back after a crash.ClaudeClaude Code
024VibeFlow-P5A p5.js particle sketch, a 356-byte Go file server and a hardened Docker image, published alongside a manifesto for vibe coding, with the conversation’s German left in the comments.Claude
020Moonshine VoiceA speech stack for voice agents and interfaces: transcription, intent recognition and text to speech, built to run at low latency on small devices, with 74 of its recent commits co-authored by Cursor.Claude
017BettaFishA public-opinion analysis system in which several kinds of research agent are made to argue with each other on purpose, so the output is not one model’s opinion written up at length.
011Open WebUIA self-hosted chat interface that talks to Ollama and to any OpenAI-compatible API.Ollama
010OllamaRuns open-weight models on your own machine and serves them on a local API.llama.cpp
009Open InterpreterA terminal coding agent that started as a local alternative to OpenAI Code Interpreter and was rebuilt in 2026 as a Rust fork of Codex, aimed at open-weight models.CodexClaude CodeDeepSeek
003ComfyUIRuns diffusion models from a node graph: connect loading, prompting, sampling, decoding and saving, then run the graph.Stable Diffusion
001whisper.cppOpenAI’s Whisper speech recognition rewritten in C/C++, so it runs on a CPU, a phone, or in a browser tab.Whisper




























































































