Harness Hub

What's new

New in the directory

What joined Harness Hub lately, which projects shipped a major version, what they have released, and what our discovery bot has spotted but nobody has reviewed yet.

Added this week

12 companies · 127 open source · 156 skills, plugins & mods · 124 events

Companies · 12

AI coding agents that let non-engineering teams ship work, and an orchestration platform for them.

A code change verification agent: risk-rated PR reviews and CI fixes that iterate until the build is green.

Hosted OpenClaw: a personal OpenClaw agent running on its own cloud VM.

An AI DevOps engineer that fixed CI failures, triaged incidents and reviewed code ("Platform engineering on autopilot").

A control plane for an AI software factory: governed, stateful agent workflows across the software lifecycle.

An agent environment from the lakeFS team: a versioned filesystem, credentials and tools for AI agents.

and 6 more entries in Companies

Open source · 127

Self-hosted workflows for orchestrating coding agents with scripts and approval gates.

An open-source terminal coding agent for large projects and real-world tasks (cloud wound down in October 2025).

Self-hosted multi-harness, multi-agent engine for teams

git worktrees + tmux windows for zero-friction parallel dev: one command per isolated agent task.

Open-source IDE platform with Theia AI, a vendor-neutral AI harness.

Open-source workflow automation with AI agents and MCP servers

and 121 more entries in Open source

Skills, plugins & mods · 156

MCP server and CLI for agents to see and control Android phones and iPhones.

Enterprise architecture governance commands for AI coding assistants

Skills for coding agents to build games and engine-rendered projects

MCP server that lets agents browse and debug web pages with Firefox

Claude Code plugin for game sprites matched to camera angle and art style

Plugin that makes coding agents follow project rules

and 150 more entries in Skills, plugins & mods

Events · 124

and 118 more entries in Events

Added in the last 30 days

217 companies · 372 open source · 360 skills, plugins & mods · 344 events

Companies · 217

AI coding agents that let non-engineering teams ship work, and an orchestration platform for them.

A code change verification agent: risk-rated PR reviews and CI fixes that iterate until the build is green.

Hosted OpenClaw: a personal OpenClaw agent running on its own cloud VM.

An AI DevOps engineer that fixed CI failures, triaged incidents and reviewed code ("Platform engineering on autopilot").

A control plane for an AI software factory: governed, stateful agent workflows across the software lifecycle.

An agent environment from the lakeFS team: a versioned filesystem, credentials and tools for AI agents.

and 211 more entries in Companies

Open source · 372

Self-hosted workflows for orchestrating coding agents with scripts and approval gates.

An open-source terminal coding agent for large projects and real-world tasks (cloud wound down in October 2025).

Self-hosted multi-harness, multi-agent engine for teams

git worktrees + tmux windows for zero-friction parallel dev: one command per isolated agent task.

Open-source IDE platform with Theia AI, a vendor-neutral AI harness.

Open-source workflow automation with AI agents and MCP servers

and 366 more entries in Open source

Skills, plugins & mods · 360

MCP server and CLI for agents to see and control Android phones and iPhones.

Enterprise architecture governance commands for AI coding assistants

Skills for coding agents to build games and engine-rendered projects

MCP server that lets agents browse and debug web pages with Firefox

Claude Code plugin for game sprites matched to camera angle and art style

Plugin that makes coding agents follow project rules

and 354 more entries in Skills, plugins & mods

Events · 344

and 338 more entries in Events

Major updates

Projects that moved to a new major version in the last 30 days, then the most-starred 0.x projects that started a new minor series (which semver allows to break things).

  1. claude-memv15.21.0 →v16.0.0Major versionPlugin

    · 3d ago

    My Work puts what needs you first My Work, the page that shows what your agents are doing, was rebuilt from a full UI review (#6210, #6304). What needs you sits at the top, long lists fold, and every page works on a phone, by keyboard and with a screen reader. What needs you…

  2. ragflowv0.27.3 →v1.0.0-rc3Major version

    · 3w ago

    1.0.0-rc3 is the third and final release candidate before the official v1.0.0 release. It fixes several issues found in rc2 and introduces new features and enhancements. Key updates: Introduced the decision model for evidence sufficiency judgment within the agentic tool loop,…

  3. OpenSpecv1.17.0 →v2.0.0Major versionMod

    · 10d ago

    OpenSpec 2.0 OpenSpec now installs skills in .agents/skills/ instead of keeping a separate copy for each coding agent. Tools that don't read this folder, like Claude Code and Kiro, get links to the same skills. We're also removing the /opsx: slash commands. Each command already…

  4. Humanizerv3.1.0 →v4.0.0Major versionSkill

    · 11d ago

    Humanizer 4.0.0 lets the skill add a specific detail where the text is generic, and it rewrites the skill's own instructions and examples to be clearer. A rewrite may now contain a detail that was not in your text. If you rely on the old rule, stay on v3.1.0. What changed The…

  5. Agentic Awesome Skills (AAS)v20.11.0 →v21.0.0Major versionSkill

    · 3d ago

    npm unavailable: This version is not published to npm; package distribution and full release alignment remain incomplete. [21.0.0] - 2026-11-18 - "50K on GitHub and 3,000+ Skills" GitHub's public counter now shows 50k stars, and Agentic Awesome Skills includes 3,022 catalog…

  6. oh-my-piv18.22.2 →v19.0.0Major version

    · 2d ago

    @oh-my-pi/pi-agent-core Added Added support for excluding custom messages from model context and compaction or branch summaries, while allowing queued steering and follow-up messages to continue correctly after excluded messages. @oh-my-pi/pi-ai Added Added a prewarm stream…

  7. Nuwa Skillv1.1.0 →v2.0.0Major versionSkill

    · 11d ago

    这一版改了什么 女娲的目标从「造出来像不像」改成「比不装 Skill 的模型多出什么」。 先写先验快照:调研之前,先把模型不查资料就知道的内容写下来。调研只找三种东西:推翻它的、比它深的、它没有的。 产物分三层:350 行以内的 SKILL.md、案例库 references/casebook.md、原话库 references/voice.md。新增招式(遇到某类问题他具体怎么做)、常见误读、边界三格,每个 Skill 自带七条运行纪律,比如「材料里没有,不等于他没说过」「原话和推演分开说」。 验证改成盲测:独立 agent 用同一组任务题,带…

  8. Parlantv3.3.2 →v4.0.2Major version

    · 2w ago

    [4.0.2] - 2026-11-05 Added The chat widget lets the AI agent in a visitor's browser talk to your agent over WebMCP (document.modelContext; in Chrome behind a flag today). On every page it's on, the widget registers an ask_agent tool described with the agent's name. A call…

  9. Phoenixarize-phoenix-v21.3.0 →arize-phoenix-v22.0.0Major version

    · yesterday

    22.0.0 (2026-11-20) MIGRATION Features add type column to generative_models (3047aab) cost-tracking: calculate costs for DECISION spans (#17764) (0f83453) cost-tracking: calculate costs for EMBEDDING spans (#17788) (0f39d64) cost-tracking: expose generative model type in…

  10. oh-my-hermesomher-v0.2.14 →omher-v1.0.0Major versionMod

    · 3d ago

    omher 1.0.0, the desktop app for your Hermes bots. If Hermes Agent is not installed yet, omher offers to install it (with oh-my-hermes) on first launch. macOS: run curl -fsSL https://omher.app/omher sh in Terminal. It installs this build with no security prompt. macOS dmg…

  11. SkillEvaluatorv0.9.0 →v1.0.0Major versionSkill

    · 4d ago

    Superseded by SkillEvaluator v1.0.1. Use v1.0.1 for the published Python package, CI bundle, and complete release assets. Original v1.0.0 release notes SkillEvaluator v1.0.0 🎉 SkillEvaluator v1.0.0 marks our first major release. Understand how your skills load, measure how…

  12. CCometixLinev1.1.2 →v2.0.0Major versionPlugin

    · 4w ago

    [2.0.0] - 2026-10-24 Added layout: Report the steps taken to fit a line layout: Wrap lines too wide for the terminal metrics: Add trend, auto-compact, run-out, spend rate and rewarm metrics metrics: Add cmd items for user commands metrics: Declare each metric's options metrics:…

  13. brooks-lintv1.9.1 →v2.0.0Major versionPlugin

    · 4d ago

    This version updates the Health, Sweep, and CI review contracts. See the migration guide below before upgrading. Breaking changes Action consumers must handle skipped=true and an empty score for an empty diff instead of treating the result as numeric zero. CI accepts review…

  14. Bernsteinv3.24.1 →v4.0.0Major version

    · yesterday

    Bernstein 4.0.0 At a glance If a workflow node fails, Bernstein skips later layers, including independent branches, as in bernstein workflow resume. A detached SSH task with a nonzero remote exit fails the run and leaves later tasks pending. bernstein govern audit --strict…

  15. unbrowsev12.5.0 →v13.0.0Major versionPlugin

    · 3d ago

    On a Mac, download the disk image: Unbrowse-macos-arm64.dmg for Apple silicon, Unbrowse-macos-x64.dmg for Intel. Open it and drag Unbrowse onto Applications: the app holds the window, the node, unb and the browser program the node fetches when a goal first needs it. It is…

Latest releases

  1. opencodexv2.100.7Mod

    · today

    Published to npm as @bitkyc08/opencodex@2.100.7 with dist-tag latest. New Features Run StepCode as a Step Plan transport (#7267) Drop Start at Login from the desktop Getting started and let Home hide it (#9007) Switch models on/off from Providers and an Overall tab in Models…

  2. Claude Code Toolsvoxtype-v0.3.20Mod

    · yesterday

    voxtype 0.3.20 — install with: uv tool install voxtype

  3. Ouroborosv0.63.0Mod

    · yesterday

    What's Changed fix(orchestrator): show the shared attempt every criterion it covers by @Q00 in https://github.com/Q00/ouroboros/pull/3073 fix(evaluation): kill mechanical command process group on timeout by @Q00 in https://github.com/Q00/ouroboros/pull/3077 feat(auto): build at…

  4. oh-my-hermesv4.0.9Mod

    · yesterday

    Codex units can write inside the fanout fence on Linux and macOS. codex exec is read-only by default, and workspace-write cannot start Codex's own sandbox inside an enforced bwrap or sandbox-exec fence, so every write a codex unit tried was refused. When the fence is enforced…

  5. CC Switchv4.0.37Mod

    · yesterday

    CC Switch v4.0.37 本版本以修复为主。读不出来的供应商和 MCP 条目(比如由更新版本写入、经 WebDAV / S3 同步过来的)不再被当成空配置、下次保存时被默认值覆盖,而是单独列出来,可以删除;Pi 新增 Mistral AI 预设;韩语界面翻译整体改进。 另外修复了本地路由监听地址填 IPv6 时启动不了;「应用到 Claude Code 插件」会把损坏的 ~/.claude/config.json 整个清空;网速慢时大的技能仓库下载到一半被中断;从访达打开 CC Switch 时 Codex 模型目录探测找不到…

  6. CLIProxyAPIv8.0.80Mod

    · yesterday

    Linux release assets CLIProxyAPI_ _linux_ .tar.gz is the default Linux build. It supports dynamic library plugins and is built against a GLIBC 2.17 baseline. CLIProxyAPI_ _linux_ _no-plugin.tar.gz is the portable Linux build for musl-based or older systems such as OpenWrt. It…

  7. claude-code-workflowsv0.31.1Mod

    · yesterday

    Work Plan tasks take their verification method and Verification Focus from any governing source they cite, such as a UI Spec, not only the Design Doc.

  8. oh-my-agentcli-v21.0.1Mod

    · yesterday

    21.0.1 (2026-11-20) Documentation skills: add cli runtime checks and patch detection to oma-qa (d059294) skills: add provenance and subtraction-first to oma-refactor (3bf310d)

  9. Codemancodeman@1.51.1Mod

    · yesterday

    Thanks @kkalisz for Send on Enter in the phone's input field (#820), codeman tui (#834), and a password-protected Codeman that installs as an app (#826). @ArizeSky for web tabs that render the DeepSeek Harness web UI and Vite dashboards (#825), and for closing a session-cookie…

  10. rulesyncv43.11.0Mod

    · yesterday

    What's Changed Features feat(musecode): generate and import Muse Code custom agents by @dyoshikawa in https://github.com/dyoshikawa/rulesync/pull/3919 feat(bob): support global subagents via ~/.bob/settings/custom_modes.yaml by @dyoshikawa in…

  11. OpenSpecv2.1.1Mod

    · 2d ago

    OpenSpec 2.1.1 This release smooths out a handful of rough edges. Skills now install straight from the CLI, with no Git and no network needed. Archive is more careful with your main specs, and validate --strict catches two problems before they reach archive. Your workflows…

  12. Spec Kitv1.4.3Mod

    · 2d ago

    Install What's Changed chore: bump version to 1.4.3 fix(extensions): report missing command files on install (#5479) test(integrations): make unknown integration checks independent of network (#5486) fix(git): handle non-ASCII branches over legacy pipes in…

  13. Failproof AIv1.0.15Mod

    · 2d ago

    failproofai 1.0.15 This release stops OpenClaw tool calls from being blocked as "failproofaid could not be reached" while the daemon is running normally. Fixes OpenClaw Tool calls are no longer blocked while the gateway is busy. The plugin timed its wait for the daemon on the…

  14. OmO (oh-my-openagent)v5.1.79Mod

    · 2d ago

    Fixed The /remote close in the TUI follows thread_close's close contract (#14234). A close refused because the thread has messages it has not read names /remote close --force and lists those messages, or says the ids are unavailable on that device's engine. The picker then…

  15. GSD Core (Get Shit Done)v1.22.0Mod

    · 2d ago

    Install What's Changed Added STATE.md size warnings — quick-tasks-append now reports a warnings entry (exit 0, row still written in full) when a task description is over 240 characters (quick-batch complete returns the same warnings for the row it appends), and /gsd-health adds…

  16. skillsharev0.30.12Mod

    · 3d ago

    skillshare v0.30.12 Release Notes TL;DR Register every Git checkout in a workspace at once with skillshare project add --scan DIR, or from the dashboard's Add project dialog. Removing a project says where its MCP and hook settings went. MCP server rows show each Agent once, and…

  17. Gentle AIv4.6.0Mod

    · 3d ago

    Gentle AI v4.6.0 is a minor release that keeps authorized work moving, makes the orchestrator responsible for what happens when a worker runs out of corrections, and improves Gentle Shell's Windows installation and uninstall paths. The review provider contract advances to…

  18. AI-companyv1.17.1Mod

    · 3d ago

    A patch release that fixes installing and updating on a Python marked as externally managed (PEP 668): Homebrew Python, and the system Python of Debian 12+ and Ubuntu 23.04+. On those interpreters a fresh python3 install.py installed the dependencies but not the aiteam package…

  19. hapiv0.36.5Mod

    · 3d ago

    What's Changed fix: align Stop with the official agent TUIs by @junmo-kim in https://github.com/tiann/hapi/pull/1992 feat(web): smooth low-rate indicator motion behind a setting by @miuiadmin in https://github.com/tiann/hapi/pull/2457 fix(web): keep chat pinned to tail when…

  20. oh-my-codex (OMX)v0.22.11Mod

    · 3d ago

    oh-my-codex v0.22.11 Summary This patch release ships two fixes for setup reliability: preserving custom developer_instructions during legacy setup mode and canonicalizing CODEX_HOME for consistent symlink handling. These changes improve setup robustness when users have…

  21. Meridianmeridian-v1.93.0Mod

    · 4d ago

    1.93.0 (2026-11-17) Features make the OAuth usage cache TTL configurable (#1502) (41bae4c) profiles: remove a Claude profile from the Profiles page (#1493) (b70a7b0) ui: mark card activation with a brand strip and show the allowance badge (#1544) (bf7835a) Bug Fixes chatgpt:…

  22. CCSv8.11.1Mod

    · 5d ago

    8.11.1 (2026-11-16) Bug Fixes update: report success when the target version lands despite a package-manager error (#1816) (77094a5), closes #1815 CI release: publish to npm via trusted publishing (#1817) (c53d92b)

  23. Mixture of Models (Claude Code mod)v0.33.1Mod

    · 5d ago

    Fixed The heartbeat continued answers that ended waiting on work running elsewhere ("Waiting for the translator to finish; then I'll merge it.", "Once it reports back I'll commit."): they announce the next step, so they read as a stop, but continuing them could only make the…

  24. Claude Code Routerv3.1.5Mod

    · 5d ago

    v3.1.5 This release contains the October/November issue fixes merged since v3.1.4. Fix provider credential fallback, vendor-qualified model routing, stale profile model selections, and context-window/reasoning metadata discovery. Preserve native Anthropic safeguards, classifier…

  25. CloudCLI (Claude Code UI)v1.37.9Mod

    · 5d ago

    1.37.9 (2026-11-16) New Features claude: offer pinned model versions in the Claude model picker (#1790) (5bf1e91) desktop: add Linux .deb build (#1092) (a8c7813), closes #1066 i18n: complete Turkish translation and add Turkish git namespace (#1768) (62cfa8e) sidebar: load older…

  26. antigravity-for-claude-codev0.31.1Mod

    · 5d ago

    Update if you are on agy 1.5.x, use background jobs (agy-job cancel), or want the Gemini 4 tier defaults. On agy 1.5.0+, an expired --print-timeout exits 3 instead of 0 and was misreported as AGY_FAILED instead of TIMEOUT (#115); agy-job cancel left timeout and agy (agy.exe)…

  27. Vibe Kanbanv0.1.46.1794744340Mod

    · 6d ago

    What's Changed bump copilot version to 1.0.83 by @anastasiya1155 in https://github.com/BloopAI/vibe-kanban/pull/3454 Add script to bump version locally by @anastasiya1155 in https://github.com/BloopAI/vibe-kanban/pull/3457 Coalesce replayed stderr chunks so stored sessions load…

  28. oh-my-claudecodev5.8.4Mod

    · 11d ago

    oh-my-claudecode v5.8.4: Hooks improvements, Jev enhancements, Teleport fix Release Notes v5.8.4 is a patch release that adds opt-in shell command checks, improves Jev model routing, fixes worktree teleport for projects without package.json, and hardens Windows plugin hook…

  29. heardv2.0.16Mod

    · 13d ago

    Heard 2.0.16 makes the texts Heard sends you shorter, and sends them only when you asked. Your phone Shorter texts. Every text Heard writes to you now goes through the same editing pass, so it leads with what you need to do and leaves the rest out. "Text me" reads the…

  30. Cavemanv4.1.0Mod

    · 13d ago

    Caveman 4.1.0 is a maintenance release: bug fixes across compress, OpenCode, Hermes, Pi, the proxy and the hooks, a recovery store that no longer fills up and stops compressing, and a few small switches people asked for. caveman-compress now refuses a file with a key or token…

From GitHub releases (or tags) of every project in the directory, checked nightly. Follow along with the Atom feed.

On our radar

UnreviewedSpotted by our discovery bot — not yet reviewed. Stars come from GitHub and are not checked; we have not tried any of these, so look before you install.

Nothing on the radar right now.

The Harness Hub newsletter

New harness companies, projects, releases and the monthly State of Harness report, in your inbox.