Skip to content

Open source AI Employees. 8 scheduled business roles, 60 routines, on Claude Code and 10 other harnesses. They drive your browser the way you do and improve every run. You own the files.

12.6 KB
Raw Blame

Harnesses

Every routine is one SKILL.md with name and description frontmatter, plain markdown instructions, and no clock time in it. Any harness that can read files, write files, run a command, read the clock, and (ideally) drive your signed in browser can run one. What differs is the scheduler and the invocation, and this page has both for each harness, with the one thing to check on a first run.

The kits are built for Claude Code, OpenClaw, Hermes, OpenCode, Grok Bot, Codex, Antigravity, Pi, Cline, Qwen Code, DeepSeek, Muse and Dots, and run on Windows, macOS and Linux. Claude Code on Windows is where the GTM Engineer has run my own club launch every weekday since August 27.

Two rules hold on every harness. Routines are scheduled jobs, not skills: point the scheduler at the kit's routines/ folder and never copy them into a global skills directory, because a skills directory loads all of them into every session you open and lets any of them be invoked outside its window, where it will only record a skip and exit. Keep one copy: every routine ends with a ## Corrections section you write into and it reads on its next run, and two copies means you write into one and it reads from the other.

Harness Status Scheduler Invocation shape What to check first Adding a connection
Claude Desktop app, Windows Ready Local scheduled tasks, one per routine, permission mode per task The task's prompt is Read <root>/routines/<id>/SKILL.md and follow it. with the working folder set to the kit Always allow on the first run of each task, Keep computer awake on A directory connector added at claude.ai/customize/connectors, which every session and every local scheduled task sees; anything else with claude mcp add --transport http <name> <url>, then /mcp to sign in
Claude Code CLI, Windows Task Scheduler Ready Task Scheduler, one task per routine pointed at run/<id>.cmd claude.exe -p "<prompt>" --permission-mode acceptEdits --output-format json < nul, shipped as run/<id>.cmd.example Full path to the binary, claude auth status, the missed start setting on The same connectors when the CLI is logged in to claude.ai, never on an API key; other servers with claude mcp add, signed in once with claude mcp login <name> before the first scheduled fire
Claude Desktop app, macOS Ready Local scheduled tasks Same as Windows Same as Windows; the macOS first run checklist is in docs/INSTALL.md Same as Windows
Claude Code CLI, macOS launchd Ready launchd, one plist per routine, StartCalendarInterval /bin/zsh -lc "claude -p ... < /dev/null", plist in docs/INSTALL.md launchd coalesces missed fires into one run on wake, which the window guard handles Same as the Windows CLI row
Claude Code CLI, Linux cron Ready cron, lines in each kit's CAPABILITIES.md 9.3 claude -p "<prompt>" < /dev/null with absolute paths cron never catches up after sleep Same as the Windows CLI row
OpenClaw Ready Built in cron: openclaw automations create "<cron>" "<message>" --name <id> --session isolated, one per routine, with the timezone flag The message is Read <root>/routines/<id>/SKILL.md and follow it. Whether its browser control attaches to your signed in profile; run the probe in CAPABILITIES.md 1.2 mcp.servers in its config, openclaw mcp login <name>; skills from ClawHub
Hermes Ready Built in cron with delivery to any platform Point each job at the routine file as its prompt Whether it reads files and runs a shell command; then the browser question mcp_servers in ~/.hermes/config.yaml, or hermes mcp install <name> from its catalog
OpenCode Ready None built in; use the operating system's opencode run "<prompt>", confirm against opencode --help Add a browser automation server for the browser lane mcp.<name> in opencode.json, then opencode mcp auth <name>
Grok Bot Ready Its recurring tasks, one per routine, on a bot named after the Employee. Every bot on your account shares one cloud computer, and the kit lives there The bot installs the kit on its own computer (npx ai-employees hire <slug> --to ~/ai-employees/<slug> in its terminal), you paste the kit's AGENTS.md into the bot's Instructions field, and each recurring task's prompt is Read <root>/routines/<id>/SKILL.md and follow it. with the fire time in the timezone SCHEDULE.md names Whether your signed in sessions are on that computer. They are only if you signed in there or you run a cookie sync to it (Agent Cookie, from a Mac over Tailscale), and every bot then shares every login. Then that the first brief arrived in the bot's thread, which is brief.deliver Its built in connectors, or a custom connector by URL at grok.com/connectors
Codex Confirmed The app's automations: one cron automation per routine, named after the routine id, local execution, the kit folder as a working directory The automation's own prompt, Shape B; codex exec "<prompt>" by hand where the CLI is signed in The sandbox and the scheduled process: confirm it can write in the kit folder, reach the network, and reach the same connections the chat session could, because the first scheduled fire is the first real test of both [mcp_servers.<name>] in ~/.codex/config.toml, then codex mcp login <name>; skills in .agents/skills
Antigravity Ready agy job runner pointed at the routine folder agy -p "<prompt>" Whether it drives your signed in browser profile or a clean one .agents/mcp_config.json in the workspace or ~/.gemini/config/mcp_config.json, with serverUrl, or its MCP Store
Pi Ready None built in; use the operating system's pi -p "<prompt>", confirm against pi --help Whether it reads the kit folder as the working directory; then the browser question None. Pi runs no MCP servers by design, so only the command line routes in a kit's section 4b apply, wrapped as skills
Cline Ready Built in cron: cline schedule create "<prompt>" --cron "<cron>", one per routine, auto approve on The prompt is Read <root>/routines/<id>/SKILL.md and follow it. That a scheduled run starts in the kit folder, and that auto approve is on so it never hangs ~/.cline/mcp.json, Streamable HTTP with a token header
Qwen Code Ready Built in scheduled tasks, or the operating system's qwen -p "<prompt>", confirm against qwen --help That it can write inside the kit folder and reach the network qwen mcp add <name> <url>, or mcpServers in ~/.qwen/settings.json
DeepSeek Ready Its scheduling plugin, one run per routine dsh runs a local server; confirm the headless prompt form against dsh --help Whether it drives your signed in browser profile or a clean one Its MCP plugin, one server per plugin row
Muse Ready Muse Code has loop jobs that stop when the terminal closes, so use the operating system's, one job per routine. Muse, the personal agent, runs crons on its own cloud machine Muse Code: muse exec with approvals disabled and Read <root>/routines/<id>/SKILL.md and follow it. as the prompt. Muse: the same prompt as a cron on its cloud machine Muse Code reads the Claude Code skill format directly and imports existing skills with one command; confirm it with muse exec on one routine. For Muse, whether its cloud machine can reach your signed in accounts at all MCP servers connected in Muse Code; connectors in Muse
Dots Ready Always on from its own cloud computer, so ask your dot to run each routine on the schedule SCHEDULE.md names, one task per routine You give your dot the kit and each routine's Read <root>/routines/<id>/SKILL.md and follow it. as the task, from ChatGPT, Slack or Teams That the dot keeps the kit folder on its own computer between runs, and can reach the accounts the routine reads. Open its computer and read the first run. Its background research uses read only tools, and sensitive actions such as changing a password always stay with you Plugins, connected through your ChatGPT app controls, with Custom Rules to allow, require approval for or block specific actions

Every kit's CAPABILITIES.md section 4b names the connections that read its accounts without a browser, with the read only form of each route. The column above is how each harness adds one. A kit never installs a connection; the install prompt checks for each and names the absent ones in its handover.

What the invocation has to get right, on every harness

  1. The working directory is the kit folder. Every path in a routine is relative to it.
  2. No permission prompt. A scheduled run in a prompting mode does not fail at 06:45; it hangs, and leaves no record. Run scheduled work in the harness's auto approve mode, scoped to the kit folder where the harness allows scoping. This does not weaken the two guardrails, which live in the contract and the routines, not in the prompt gate. CAPABILITIES.md section 10 in every kit is the full argument.
  3. Prove one routine by hand before registering the rest. Run the standup, watch it write brief-latest.md and exactly one line into runlog.jsonl. Eight jobs registered on an invocation nobody ran is eight silent failures on the same morning.
  4. A non zero exit should leave a record. The Windows launcher does this through runlog.mjs --failed-run. On other harnesses, add the same || fallback to the command line.

Grok Bot, and the hosted shape it set

Grok Bot is the one harness on this page whose computer is not yours, and the shape is worth writing down once because Codex cloud, Meta Muse and OpenAI Dots share it. Everything here comes from operators' published accounts of the product in August 2026, and each kit's CAPABILITIES.md marks it expected until you write confirmed into ## Corrections.

  • One computer for every bot. All the bots on an account share one persistent Linux machine with a terminal, file access and a real browser. Each bot has its own screen and runs one computer-use task at a time; several bots can drive the browser at once. So the kit lives on that computer, the bot installs it there itself, and the cloud sync rule does not apply.
  • One bot per Employee, one task per routine. Name the bot after the role, paste the kit's AGENTS.md into its Instructions field (it is the map that points at the files, and it is short enough to fit), and register one recurring task per routine with the Shape B prompt. Do not turn the routines into the bot's workflows: a workflow is invoked on demand, a routine runs in a window.
  • Bots are not a security boundary. Files, browser sessions and app logins are shared by every bot on the computer. A release you write for one Employee is a file every bot can read; a session you sync there is a session every bot can use. Scope by what you sign in to, never by which bot you talk to, and start read only on public pages before you sync anything.
  • Staying signed in is a sync you run. Nothing in a kit signs in, so your sessions reach the bot's browser only if you signed in there or you keep them there with a cookie sync such as Agent Cookie (Chrome cookies from a Mac to the bot over Tailscale, every fifteen minutes). The kit names it under browser.session, detects it, and installs nothing.
  • Dots is the same shape inside ChatGPT. OpenAI announced it on September 29, 2026: each dot has its own cloud computer and browser, is powered by GPT-6 Astra, connects to apps through plugins, and reaches you from ChatGPT, Slack and Teams. It is rolling out to Pro and Business Premium plans in eligible markets, with an Enterprise beta. You can open its computer at any time, which makes the first run easy to inspect. Its Activity View shows background work, and Custom Rules set what it may do without asking.
  • The brief comes to your thread. You never open that computer's files, so the standup resolves brief.deliver and posts brief-latest.md into the bot's own thread, which the Grok app carries to your phone. The file stays the record.
  • Two more routes. Claude Code can be logged in on the bot's computer, which turns the Grok Bot row into the Claude Code CLI row with Grok Bot as the scheduler. Peekaboo, installed by you on a Mac, lets the bot see and click native Mac apps over the same link. Both are yours to add.
  • Watch the meter. Plans carry a weekly usage allowance and fleet work moves it fast. Start on the lowest tier that includes bots, run one Employee for a week, and read the meter before you add a second.

Ran a kit on one of these?

Open an issue with the harness and version, the invocation you used, what the first run record said, and what you changed. It goes into the notes for that row, with your name on the change.