This is a very opinionated list from my daily usage. I will try to keep 🔥 entries below 10 (or 20, if I can't 😉).
See AGENTS.md for contribution guidelines when adding entries.
Visit the web page version here. It is generated directly from the catalog tables in this README plus deployment-time GitHub activity metrics.
| Icon | Meaning |
|---|---|
| 🔥 | Long-term daily driver — proven, highly recommended |
| 🧪 | Actively trying out, but not daily used |
| 👀 | Recently discovered, looks promising, not yet tried |
| Status | Tool | Tags | Description |
|---|---|---|---|
| 🔥 | oh-my-pi | hash-edits, lsp, browser, subagents | AI coding agent for the terminal — hash-anchored edits, optimized tool harness, LSP, Python, browser, subagents, and more |
| 🧪 | Claude Code | terminal, codebase, git-workflows | Agentic coding tool that lives in your terminal, understands your codebase, and helps you code faster by executing routine tasks, explaining complex code, and handling git workflows — all through natural language commands |
| 🧪 | Codex CLI | lightweight, terminal, cli | Lightweight coding agent that runs in your terminal |
| 🧪 | 3code | cost-optimization, caching, compaction, byo-provider | The economical coding agent — token budget as a first-class constraint (chunked context, supersede-aware compaction, aggressive caching); any OpenAI-compatible endpoint, 75% fewer tokens than OpenCode on a SWE-bench subset |
| 👀 | molt | verification, receipts, desktop, byo-model | A coding agent that won't say done on a false claim. Verification on disk. Receipts for accepts and refusals. Terminal and desktop. OpenAI compatible or Anthropic. |
| 👀 | mu | judge-model, decision-ledger, pi, desktop | mu (μ): a coding agent that thinks before it acts. A small, fast judge makes the routine calls, the big model does the work. Built on pi and AionUi. |
| 👀 | Mixdog | desktop, model-routing, subagents, multi-provider | Open-source desktop coding agent where a lead model plans and cheaper models handle search, edits and review. |
| Status | Tool | Tags | Description |
|---|---|---|---|
| 🔥 | sandboxed-agents | bwrap, sandbox, profiles | Unified bwrap sandbox wrapper for multiple coding agents — oh-my-pi, Claude Code, Codex — with per-agent profiles and layered config |
| 👀 | codex-profiles | codex, profiles, account-switching | Codex CLI/Desktop profile switcher that launches Codex with isolated CODEX_HOME directories for separate auth, config, sessions, plugins, and logs |
| 👀 | SandBase CLI | mcp, model-routing, cli | Local MCP bridge CLI that lets coding clients discover and run 2,000+ AI models and APIs |
| 👀 | SandBase Harness | mcp, sandbox, sessions, self-hosted | Local-first, self-hosted AI agent runtime and MCP bridge with sandboxed sessions, memory, credentials, audit/replay, and a local Console |
| 👀 | codex-fixes | codex, fixes, issue-registry | Checks your local Codex install for known issues and runs the matching community-maintained fix scripts |
| 👀 | codex-plan-ledger | codex, plan-mode, decisions, drift-check | Decision ledger for Codex /plan: Codex's own questions and answers, open decisions and plan defaults become a reviewable, diffable decisions.json in your repo, plus a scope drift check after coding. |
| 👀 | Ariadne Loop | loop-engineering, verification, agent-packets | Local-first CLI and templates for turning rough coding-agent context into verifiable loop specs, agent packets, verifier gates, stop rules, rollback rules, and JSON reports |
| 👀 | py_ralph_frame | loop-engineering, verification, agent-loops | A lightweight spec-driven loop harness for Claude Code, Codex, and Gemini CLI. Fresh-context runs, specs, commits, and gates to get you and your agents productive immediately. |
| 👀 | forkd | microvm, fan-out, snapshots | MicroVM sandbox runtime for AI agent fan-out — fork 100 microVMs in ~100ms, BRANCH a live VM in ~150ms, KVM-isolated snapshot CoW |
| 👀 | Agent Teams | desktop, orchestration, kanban, review | You're the boss, agents are your team. They handle tasks on their own, message each other, and review each other's work. You just watch the kanban board and give high-level commands. Codex/Claude/OpenCode(200+ models, 75+ LLM providers, free models no auth). Build your AI company with multiple teams. |
| 👀 | Agentlas OS | agent-os, orchestration, local-first, multi-model | Agent OS: keep specialist agents in a hub, spin up a temporary orchestrator per task. Local-first, works with any model. |
| 👀 | Tree Ring Memory | memory, local-first, sqlite, tui | Framework-agnostic, local-first memory lifecycle for AI agents: Rust CLI, SQLite/FTS recall, forgetting, audit, consolidation, DOX/Revolve adapters, and TUI. |
| 👀 | Oh My Hermes | hermes, skills, workflows | Workflow skill pack for Hermes Agent that adds research, planning, coding handoff, review, QA, documentation, and long-running loop workflows |
| 👀 | pi-custom-packages | pi, extension-pack, wsl, screenshot, session-guards | Ten extensions for the pi coding agent: session guards, image context control, and Windows/WSL screen capture, cursor, and recording. |
| 👀 | Agent Island | macos, windows, status, local-first | Open-source status companion for Claude Code, Codex, Antigravity, Grok and Cursor with live session state, your-turn alerts, and locally computed usage and cost |
| 👀 | termagitchi | status-line, worktrees, gacha, git-hygiene | A creature for every git worktree, living in your coding agent's status line. Deterministic species, rarity bands, and one view across every live session. |
| 👀 | agent-watch | monitoring, stall-detection, preflight, posix-sh | Monitor Claude Code and Codex CLI agents for running, completed, failed, or stalled jobs. |
| 👀 | Superagent | macos, browser, ios-simulator, phone-relay | Desktop app giving Claude Code and Codex a real browser to drive, an iOS Simulator to install and screenshot apps in, and a phone companion app for remote monitoring |
| 👀 | RDLeader | agent-ops, control-plane, approvals, qa-evidence | Local-first control plane for supervising AI R&D workers — task ownership, runtime dispatch, result collection, approval gates, public-safe demo reset, and QA evidence |
| 👀 | OpenAgentRelay | trusted-lan, delegation, json, local-first | Turn any local agent or automation into a team-callable capability. |
| 👀 | breaking-coding-chaos | control-plane, skills, human-in-the-loop | Human-in-the-loop dual-loop skill suite for coding agents: throughline progress on disk, plan-spar alignment, then minimal clean-cut implement |
| 👀 | Agent-Ready Repo Auditor | repository-audit, agent-readiness, github-actions | Free GitHub Action that scores public repositories for Codex, Claude Code, Copilot, and Cursor readiness—without cloning or executing code. |
| 👀 | Better Agent | orchestration, dashboard, approvals, multi-provider | Local web workspace for coding agents — Claude, Codex, Gemini, and more in one inspectable, LAN-ready hub. |
| 👀 | Claudexor | control-plane, multi-harness, best-of-n, review | Local-first control plane for AI coding harnesses (Codex, Claude Code, Cursor, OpenCode): best-of-N runs, multi-model review, evidence-driven delivery |
| 👀 | fractal | orchestration, recursive-delegation, git-worktrees, tui | Hierarchical agent loops with recursive self-organization. |
| 👀 | aGiTrack | git, commits, token-accounting, dashboard | Terminal wrapper for Claude Code, Codex, and OpenCode that commits each agent turn to git, recording the prompt, model, and that turn's input/output/cache token counts in the commit message, with a local dashboard over the resulting history |
| 👀 | Sillage | web-ui, self-hosted, pwa | Mobile-first self-hosted web UI that drives the native Claude Code and Codex CLIs on your own machine; sessions that outlive the client, an IDE panel, its own MCP server, single Docker container |
| 👀 | Claudette | mobile, control-plane, pty, ios, android | Native iOS/Android/Apple TV mobile control plane for local CLI coding-agent sessions: real PTY plus context/cost gauge, agent tree, voice, and Take the Wheel handoff; companion CLI npx claudette setup |
| 👀 | Chump | multi-agent, coordination, local-first, offline | Self-hosted AI coding agent with persistent memory and bounded autonomy. Local-first, your keys, your data. Written in Rust. |
| 👀 | pisesh | tui, sessions, favorites, search | Keyboard-driven TUI to bookmark, search, rename, and resume pi coding-agent sessions — per-project view, cwd overrides, optional AI title generation, zero dependencies |
| 👀 | AI Badger | local-first, context, code-review, handoff | AI Badger - Local-first tool that extracts focused repo context for any AI chat (Claude, ChatGPT, Grok, etc.) without wasting tokens on irrelevant files. |
| 👀 | kgai | memory, decisions, local-first, s3-sync | Shared memory for AI dev teams — an immutable knowledge graph of the decisions behind your code, auto-captured by your agent and synced without merge conflicts. |
| 👀 | YYLO | orchestration, kanban, git-worktrees, multi-agent | YYLO (why-lo): AI coding-agent orchestration CLI with equivalent yylo and yy launchers. |
| 👀 | YYLO Benchmark | evaluation, benchmarks, evidence, worktrees | YYLO Benchmark: longitudinal evaluation and immutable evidence for agent runs. |
| 👀 | Tintpad | macos, launcher, hotkey, local-first | It falls out of your notch. Hotkey, repo, Return: your terminal opens with Claude Code, Codex or any agent already running. Free, MIT, local-only macOS. |
| 👀 | Orkas | desktop, orchestration, local-first, coding-agents | Orkas is an open-source, local-first AI desktop app: a commander LLM directs specialist sub-agents, and runs your installed coding CLIs — Claude Code, Codex, OpenCode, OpenClaw, Hermes — as local sessions. Agents self-evolve via reflection and skill crystallization. BYO keys. macOS / Windows / Linux. |
| 👀 | check-docs | agents-md, claude-md, pre-commit, stale-paths | A small sh pre-commit check that stops a commit when CLAUDE.md, AGENTS.md or the files they @import name a path that no longer exists or a name you deleted |
| 👀 | OrcaReplay | replay, tracing, proxy, offline | Time travel for AI agents — record, replay, fork, and debug any agent run with any model |
| 👀 | swe-mux | multi-agent, status, mobile, tailscale, pty | Terminal multiplexer and agentic control plane with tailscale enabled. optimized for mobile development |
| 👀 | Coven | daemon, pty, project-scope, multi-harness, sqlite | Local-first runtime for project-scoped AI coding-agent sessions, with durable state, authority boundaries, and multi-harness interoperability. |
| 👀 | Psyche Build | desktop, git-worktrees, tmux, parallel-agents | Desktop cockpit for running parallel AI coding agents in visible, isolated workspaces — from branch to review, merge, and handoff. |
| 👀 | Ordewell | orchestration, plan-first, multi-harness, verification | Multi-agent task orchestration for coding agents. Turn one goal into an ordered plan of tasks — each with its own runner, model and mode — then execute and verify the results. |
| 👀 | oxi | desktop, local-models, acp, native | Native, local-first coding-agent desktop app in Rust (egui) — run any model: local GGUF, Ollama/LM Studio, SSH-tunneled runtimes, hosted APIs, or Claude Code / Cursor / Codex over ACP. No Electron, no cloud lock-in. |
| 👀 | rerun-bench | evaluation, reruns, pass-k, cost-spread | Same task, run N times: how consistent and how expensive is your coding agent? A rerun benchmark for Claude Code, Codex and OpenCode. |
| 👀 | open agent lab | evaluation, results-site, confidence-intervals | Independent, reproducible evaluation of coding agents: same tasks, many runs, intervals shown. |
| 👀 | snap-back | undo, snapshots, shadow-git, hooks | Undo for any coding agent. Snapshots your project in a shadow git repo so one command rolls back what the agent did, without touching your own git history. |
| 👀 | agent-fence | permissions, policy, hooks, audit-log | One permission policy file for every coding agent: allow, ask or deny shell commands, file access and git operations, with an audit log. |
| 👀 | Lockpaw | macos, menubar, hooks, notifications | Covers your Mac's screen with a hotkey while agents keep running, lid closed too. The locked screen glows when Claude Code, Codex, Gemini CLI, Cursor, Copilot CLI or Aider needs you. Free, MIT. |
| 👀 | Tale | sandbox, orchestration, kanban, self-hosted | Open-source project workspace for teams and AI agents. Assign tasks, coordinate agents in persistent sandboxes, and review results together. Self-hosted or managed cloud. MIT licensed. |
| 👀 | Helicon | muse-code, desktop, vscode, diffs, approvals | Open-source desktop app and VS Code extension for Meta's Muse Code CLI, with threads, inline diffs and approvals on Windows, macOS and Linux. Unofficial, not affiliated with Meta. |
| 👀 | Oh My Android | mcp, android-emulator, adb, macos | Android emulator & ADB GUI for Mac: dark mode, font size, RTL, TalkBack, network, GPS, Layout Inspector in dp. Built-in MCP server lets Claude Code, Codex and Cursor drive the emulator. |
| 👀 | subagent-tuner | claude-code, subagents, model-selection, effort | Claude Code mod: picks the model and reasoning effort of every subagent |
| 👀 | harness-scope | claude-code, profiles, per-repo, skills | A Claude Code mod that turns your global skills, agents, rules and tools on or off per repo with named profiles. In my writing repo, it hid 94 of the 101 skills with one line. |
| 👀 | SpaceApp | self-hosted, web-ui, multi-agent, docker, source-available | Self-hosted web workspace that runs OpenCode, Codex, Claude Code, Gemini CLI and other coding CLIs side by side in persistent rooms, with MCP servers and managed browser panes |
| 👀 | CurlShot MCP | mcp, screenshot, pdf, hosted | Hosted MCP server for CurlShot: let AI agents take screenshots of any web page (PNG, JPEG, WebP, PDF, video). Setup for Claude, Cursor, VS Code. |
| 👀 | Sqwakie | multi-agent, peer-to-peer, kanban, mcp, source-available | A walkie-talkie for your team's AI agents: Mission Control, project boards with Data Rooms, seats, peer-to-peer Walkie Direct, pooled local LLMs. FSL-1.1-ALv2. |
| 👀 | Alchemist Coder | desktop, acp, subagents, local-models | Free, open-source desktop app for Claude Code and Codex: follow every agent and subagent live, run your own agent organization and automate routines. Also searches your Gemini CLI and Grok history. macOS, Windows, Linux. |
| Status | Tool | Tags | Description |
|---|---|---|---|
| 🔥 | coding-agent-instructions | instructions, rules, progressive-disclosure | Rules for coding agents. Modular design, progressive disclosure without eating your context window or distract your agents |
| 👀 | Agent Quality Kit | agents-md, quality-gates, exit-codes, compliance-levels, zero-dependencies | Turns the rules an agent is supposed to follow into commands with exit codes, and reports which of them actually run. Levels AQK-0…3 computed by a run. Zero dependencies. |
| 👀 | AGENTS.md Starter Kit | agents-md, templates, cli, github-actions | One-command AGENTS.md starter templates for monorepos, Python, Next.js, Codex, Claude Code, Cursor, Copilot, Gemini CLI, and Windsurf. |
| 👀 | toldya | claude-md, agents-md, local-first, zero-dependencies | Stop repeating yourself to your AI. Finds what you keep telling your coding agent, writes it into CLAUDE.md / AGENTS.md, counts whether it stopped. |
| 👀 | since-cutoff | agents-md, knowledge-cutoff, api-changes, type-checked, python | Find which of your exact dependency versions your coding agent writes wrong, and fix it with a small AGENTS.md note verified by a type checker. |
| Status | Tool | Tags | Description |
|---|---|---|---|
| 👀 | wiki | knowledge-base, markdown, cli, agent-skills | Indexed knowledge bases with command-line tools for agents. |
| 👀 | chamnan | context, architecture-index, impact-map, local-first, stdlib-only | Preserves a long-lived repository's engineering context — an architecture index, an impact map, session records, and the decisions behind them — as markdown committed beside the code, so an agent reads instead of rediscovering. |
| 👀 | ContextStream | context, mcp, coding-agents | Shared project context for Cursor, Claude Code, Codex, Grok. Site https://contextstream.io. MCP https://mcp.contextstream.io/mcp. Intelligence isn’t the bottleneck. Context is. |
| 👀 | Bhawna Skills | decision-memory, architecture-guardrails, invariants, cli, local-first | Decision memory and architecture guardrails for AI coding agents. |
| 👀 | Hyperconsciousness | memory, encrypted, mcp, rust | iCloud for agents. An encrypted, permissioned knowledge layer for humans and agents. |
| 👀 | Archcore | context-engineering, spec-driven, mcp, git-native | Spec-driven development and context engineering for Claude Code, Cursor, Codex, and GitHub Copilot — backed by project context in Git. |
| Status | Tool | Tags | Description |
|---|---|---|---|
| 🧪 | rtk | token-saving, proxy, rust | CLI proxy that reduces LLM token consumption by 60-90% on common dev commands — single Rust binary, zero dependencies |
| 🧪 | ponytail | token-saving, agent-skills, minimalism | Makes your AI agent think like the laziest senior dev in the room. The best code is the code you never wrote. |
| 👀 | agent-scripts | token-saving, build-output, local-first | Token-efficient, local-first CLI tools for coding agents - compact Maven, npm/Node, and Go test output plus reusable development helpers. |
| 👀 | ctxtrim | context, token-saving, cli | Trim what bloats your AI coding context — find the files ballooning your Claude Code / Cursor / Codex token cost and write ignore files to cut it. Zero-dep. npx ctxtrim |
| Status | Tool | Tags | Description |
|---|---|---|---|
| 🧪 | OpenUsage | subscriptions, usage, dashboard | Track all your AI coding subscriptions in one place |
| 👀 | ax | telemetry, dashboard, mcp, usage | Local-first observability and memory layer for AI coding-agent sessions |
| 👀 | Relay | llm-proxy, provider-switch, usage, quota-rotation | Universal AI API proxy — hot-switch between GitHub Copilot, OpenAI, Anthropic, DeepSeek, Groq, Ollama, and 10+ providers from Claude Code, Codex CLI, Aider, or any coding agent without restarting. Auto-rotates accounts when quota exhausted, real-time usage tracking |
| 👀 | Usage HUD | macos, menubar, usage, confidence | Usage HUD: one macOS menu-bar meter for Codex, Claude, Gemini, Grok and local models, with a confidence label on every number |
| 👀 | nerf-watch | session-logs, regressions, local-first | Find out when your coding agent quietly got worse or more expensive. Local-first checks on Claude Code and Codex session logs, no telemetry. |
| 👀 | tokenreceipt | usage, cost, local-first, receipt | A till receipt for your AI agent token usage at API list prices. Claude Code, Codex, OpenCode, Gemini CLI, Cline, Roo, Kilo, Pi, Hermes. Local only. |
| 👀 | NextReset | codex, reset-history, incidents, local-timer | Independent Codex reset-history tracker with official incident sources and a browser-local personal timer; historical observations do not guarantee future reset timing |
make build
make run
make deploymake buildwritesdist/index.html.make runserves the generated site locally on127.0.0.1:8080; override withmake run HOST=0.0.0.0 PORT=3000.make deployvalidates, tests, and builds the Pages artifact.npm run validate:catalogvalidates the README catalog tables.
Catalog entries live directly in README tables with four columns: Status, Tool, Tags, Description. Tool names are Markdown links to GitHub repositories, and any ## section with that exact table header is treated as a category.
GitHub metrics are fetched at build time and are never committed back into README. Tags should highlight distinguishing traits, such as microvm, sandbox, token-saving, browser, or subagents.
To add or update entries with an agent, use .agents/skills/add-tool/SKILL.md; Claude-compatible setups can use .claude/skills/add-tool.