Skip to content

Now hiring · AI employees, new in Foreman 1.1

An AI software engineering team that works by your rules

Foreman is a free, open-source desktop app for macOS and Windows that runs Claude Code, Codex, Cursor and OpenCode in parallel, each in its own git worktree on its own branch. Its AI employees turn that into a team: roles in plain words, a hard hat each, a shared memory, and access you approve.

macOS & WindowsApache-2.0No accountYour own agent logins

An organization's Team page: its employees, what each needs from you, and a memory change waiting for approval
The real Team page: a column per employee, what each needs from you, and a memory change waiting for approval.

The role

What is an AI employee in Foreman?

An AI employee is a standing member of an organization: a name, a role in plain words, a hard hat in its own colour, instructions, memory and access to some of the organization's repositories.

It works when you ask, on a schedule, or when an event or a colleague hands it a task. Each attempt is an ordinary Foreman workspace and session, so nothing new runs underneath: same worktrees, same diffs, same approvals.

  1. A role in plain words. “Staff QA engineer: reviews what others build and says what breaks.”
  2. A hat in its own colour. Twelve colours. The hat says who; it never says state.
  3. Instructions in three layers. The organization's, the job's own prompt, then the task.
  4. Access you approve. None, Read or Write per repository; no shell unless you say so.

Engineering · Template

Staff QA engineer

Comes as Quinn · green hat

Reviews what others build and says what breaks.

Responsibilities

  • Check each change against what was asked, item by item
  • Find what breaks at the edges and what the tests don't cover
  • Write test plans someone can run
  • Report each problem with where it is, how to show it and a fix
Access
Write on the organization's repositories · no shell · Slack and Linear held
Starts
When you ask · plus any schedule or event you add and approve
Memory
memory/bots/e-9q4c/ in the memory repository
Runs on
Claude Code, unattended, in a fresh worktree each run

Try it here

Build your team

Hire from the fifteen job templates Foreman ships. Each one arrives approved, with its job's own prompt, a free hat colour and the same access. Then ask Priya for something and watch the work come back to you.

Open roles 15

Engineering

Design

Product

Data

Writing

Marketing

Customers

Slack posts and Linear changes start held for you. Since v1.2.4 a template employee pushes and opens draft pull requests on its own; this switch is the one-line rule that holds those too.

Your team 5

  • You

    The person · one of the team

    Approve each plan and each hire, and anything you've asked the team to hold.

  • Priya Lead

    Program manager

    Plans every project, proposes who it needs, reviews each part and accepts the result.

    Starts
    When you @Priya
    Access
    Writes project filesNo shellSlack · held
  • Ravi

    Staff software engineer

    Builds and fixes features, with tests, on its own branch.

    Starts
    When you askWhen assigned in Linear
    Access
    Write · atlasNo shellPush & PR · heldSlack & Linear · held
  • Quinn

    Staff QA engineer

    Reviews what others build and says what breaks.

    Starts
    When you askWhen review is requested
    Access
    Write · atlasNo shellPush & PR · heldSlack & Linear · held
  • Uma

    Staff UX designer

    Reviews screens and flows, specifies what to build and draws it as mocks.

    Starts
    When you askWhen review is requested
    Access
    Write · atlasNo shellPush & PR · heldSlack & Linear · held

acme Acme's channel · atlas

  1. You

    @Priya add passkey sign-in to atlas

  2. Priya

    I read atlas first: sign-in lives in src/auth/. Two questions before I plan.

    1. Keep passwords as a fallback? Why: decides whether existing accounts move.

    2. Must Safari work on day one? Why: Safari's passkey sheet needs its own test.

  3. You

    Keep both. Yes, Safari too.

  4. Priya

    Plan v1 · small · 3 parts

    Done when you can sign in to atlas with a passkey in Chrome and Safari, and passwords still work.

    1. 1Spec the passkey screens, two mock optionsUmaspec.md
    2. 2Test plan for Chrome and SafariQuinntest-plan.md
    3. 3Build it, run the plan, pushRavibuild.md
  5. You approved plan v1 · Uma, Quinn and Ravi start in order
  6. Ravi

    Built passkey sign-in on passkey-sign-in from Uma's option B. Ran Quinn's plan: 18 of 18 cases pass in Chrome; Safari needs you to try the sheet by hand. Asked to push.

  7. Needs you · Push · Ravi · just now

    Push passkey-sign-in to atlas

    3 commits ahead of main · 6 files · origin github.com/acme/atlas

    Push branch sends 3 commits to origin. It leaves this computer and can't be undone.

    Don't pushPush branch

Start a project with @Priya, or ask one person with @Name. Try one:

Illustrative. In the app, Priya's plans, questions and held actions read the same way.

Working hours

When does an AI employee start work?

When you ask, when a schedule fires, when an event happens, or when a colleague hands it a task. Every one of those becomes a task first, then a run in a fresh workspace.

  • You ask

    Type @Ravi … in the organization's channel, or press Ask on Ravi's page. @Priya … starts a whole project.

    @Ravi fix the flaky auth test

  • A schedule

    Cron in your time zone, an interval, or once, hourly, daily, weekly or monthly. Never more often than every 15 minutes. Missed while Foreman was closed? Skip it, or run it once.

    cron 0 9 * * 1-5every 90mmissed: run-once

  • An event

    A pull request merges, fails CI, or gets a review request, changes requested or an approval. The base branch moves. A memory file changes. A Linear issue is assigned.

    When CI fails on a PRWhen assigned in Linear

  • A hand-off

    An employee may delegate to the colleagues you allow. Chains stop at depth 2, three hand-offs a run and 20 runs a root.

    Ravi → Quinn: review build.md

Clocking in without you

  • Schedules and events run unattended, so they run on Claude Code. An approval one hits is denied and the turn stops.
  • They run while Foreman runs, with or without a window, or from login. Pull requests are checked about every 2 minutes, Linear every 5.
  • A new hire runs only when asked. Triggers are yours to add, and you approve each one with the definition.
The Automations library of templates and commands, with schedules beside it
The real Automations page: the seven built-in templates, and Max's Monday schedule on the right.

Institutional memory

How does an AI employee remember?

Through the organization's memory repository: a private git repository where each run's notes land as one commit, so git log says who learned what, and when.

  • Every run starts a fresh conversation. Memory, not chat history, is what carries over.
  • A run reads PROJECT.md, its own MEMORY.md and last RECAP.md, and searches the rest.
  • Commits carry Foreman-Employee, Foreman-Run and Foreman-Task trailers, so blame works too.
  • Everyone reads everything; each writes only its own folder. Shared files are proposals you keep or discard.
  • Memory is data, never instructions. Secret-shaped values are refused.

acme-memory

git log --format='%h %an %s' -- memory/

  1. c41f9e2Quinnlearned the test DB must be reset before e2ememory/bots/e-9q4c/MEMORY.md
  2. 7d02b11Youkept Uma's proposal: every screen works at 390pxmemory/org/PROJECT.md · Foreman-Proposed-By: e-4ux8
  3. 5be8a90Ravirecap: flaky auth test fixed; Safari cookie check leftmemory/bots/e-3fq9/RECAP.md
  4. 19ad3f4Priyaplan v2 for checkout-redesignmemory/org/projects/checkout-redesign/plan.md
acme-memory/
  memory/org/
    PROJECT.md        # pinned · proposals only
    decisions.md      # append-only
    projects/         # Priya's plans, each part's file
  memory/bots/e-3fq9/
    MEMORY.md         # Ravi's own facts
    journal.md        # one entry per run
    RECAP.md          # read first next run
  bots/e-3fq9.toml    # Ravi's definition

House rules

What stops an AI employee from pushing on its own?

A rule per outward action: allowed, held or never. A held push, pull request, Slack post or Linear change waits in Needs you as a card that shows exactly what goes where, until you press its button.

Slack posts and Linear changes start held. Since v1.2.4, pushes and draft pull requests start allowed so employees can finish; one line in the definition holds them too.

  • 01

    Nothing that leaves your computer has a keyboard shortcut. Every held action is a button named for the act: Push branch, Post to #launches.

  • 02

    Nobody approves for you. Unattended runs are Claude-only today, and any approval one hits is denied and the turn stops.

  • 03

    No employee answers another's approval, gate or question. A question is never a permission.

  • 04

    No shell by default. Foreman's own tools only; a shell is a separate approval that says plainly it acts with no hold.

  • 05

    Protected paths stay protected. .claude/, .foreman/, .mcp.json and .git/ are denied to every employee run.

  • 06

    A changed definition pauses the employee until you approve it again, with each difference shown old → new.

Benefits

What does every AI employee get on day one?

A desk, a schedule, a way of working, a toolbox, a radio and an off-site trailer. An employee's run is an ordinary Foreman workspace, so it gets everything yours do.

  • A desk of their own

    One worktree per run

    Each run is a fresh git worktree on its own branch. You read the change as a diff before it goes anywhere, and your main checkout stays untouched.

    A project's runs share one workspace, so the tester reads exactly what the builder left.

  • A schedule

    The Automations library

    Employee schedules sit beside yours, with templates to run now or on a schedule. Two opt-in helpers watch your work: auto-fix sends failing CI back to the agent, two fixes a commit and five a PR, and auto-review has a read-only reviewer read each finished diff.

    • Explain code
    • Fix a bug
    • Fix an issue
    • Write tests
    • Refactor
    • Review my changes
    • Set up this repo
  • A way of working

    Priya plans, flows repeat

    Tag @Priya and she plans the project. For a way of working you repeat, a flow loops until the checks pass, with gates where you decide. 11 ship built in.

    Experimental Flows stay off until you turn them on.

  • A toolbox

    Linear, GitHub, Slack

    A Linear assignment starts work. GitHub pull request events start work, and employees push and open draft PRs. Slack rules notify you or a channel. Langfuse traces, read-only.

    Slack is outbound only: you can't answer from it. Slack posts and Linear changes wait for you.

  • A radio

    Answer from your phone

    Pair a phone to follow the team live and answer your own workspaces' approvals, questions and flow gates, over your own Tailscale network when you're out.

    The phone can't answer employees' held actions. Those wait in Needs you, on your computer or in a browser with Full control.

  • An off-site trailer

    E2B, Daytona, Vercel, Docker

    A workspace can run in a sandbox on your own E2B, Daytona, Vercel Sandbox or Docker account. Provider keys stay on your computer.

    Experimental Organization-wide sandbox defaults for employees aren't built yet.

Other offers

How is Foreman different from Conductor and T3 Code?

Foreman is Apache-2.0 and runs on macOS and Windows, and of the three it alone lists AI employees with a git-backed memory, flows that loop until checks pass, and a CLI where every command takes --json.

A dash means not listed on their site

Foreman compared with Conductor and T3 Code
FeatureForemanConductorT3 Code
Open sourceApache-2.0NoMIT
macOSYesYesYes
WindowsYesMac onlyYes
One git worktree per agentYesYesYes
AgentsClaude Code, Codex, Cursor, OpenCode, any ACP agentClaude Code, Codex, CursorClaude Code, Codex and more
Flows that loop until checks pass, with human gatesYes——
AI employees with git-backed memoryYes——
Scheduled automations and auto-fix CIYes——
Your own cloud sandboxes (E2B, Daytona, Vercel, Docker)Experimental——
Scriptable CLI, --json on every commandYes——
Approve from your phoneYes—Yes
No account, no telemetry uploadYes——
From each project's own site as of November 2026. “—” means it isn't listed there, not that it's missing. Spot a mistake? Tell us.

Onboarding

Get started in a minute

  1. Download Foreman for macOS or Windows. Free, no account. You need Git on your PATH.
  2. Add a git repository and sign in an agent in Settings → Agents, with your own Claude Code, Codex, Cursor or OpenCode login.
  3. Create workspaces, one per task. For a team, open Personal ▾ → New organization…, pick some jobs, and say hello to Priya.

Or from a terminal

Start a workspace from a template and wait for the turn

foreman run fix-a-bug "the login page loops after sign-in" --repo atlas --wait

See what is waiting for you

foreman approvals

Open a draft pull request

foreman pr create atlas --draft

Automations from a terminal: foreman schedules, foreman flow run and foreman pr auto-fix. CLI reference

FAQ

Questions developers ask

What is Foreman?

Foreman is a free, open-source desktop app for macOS and Windows that runs AI coding agents in parallel. Claude Code, Codex, Cursor and OpenCode each work in their own git worktree on their own branch, while you review diffs, answer questions and open pull requests from one window.

Is Foreman free and open source?

Yes. Foreman is free and licensed under Apache-2.0, with the full source on GitHub at github.com/elite-build/foreman. There is no paid tier, no account and no Foreman cloud service.

Which coding agents does Foreman support?

Foreman runs Claude Code, Codex, Cursor and OpenCode, plus any agent that speaks the Agent Client Protocol (ACP). Each agent signs in with your own subscription or API key, set in Settings → Agents.

How do I run multiple Claude Code agents in parallel?

Add a git repository to Foreman and create one workspace per task. Each workspace is a separate git worktree on its own branch, so several Claude Code sessions edit the same repository at once without conflicts. Tiles shows them side by side.

Why does each agent need its own git worktree?

A git worktree is a second checkout of the same repository on a different branch. Giving each agent its own worktree means agents never overwrite each other's files, every change lands on a reviewable branch, and your main checkout stays untouched.

Does my code leave my computer?

Only through the agents you choose. Foreman runs locally with no account and no telemetry upload; your agents talk to their own model providers. In your own workspaces, an approval marks any command that leaves your computer before it runs, and AI employees hold Slack posts and Linear changes for you.

Do coding agents need my approval for every command?

Only if you want them to. Approvals are a setting: with Claude Code tool approvals on (Settings → Account), each command is spelled out part by part before it runs, anything that leaves your computer is marked, and Always allow shows exactly what it adds. Every agent still works in its own worktree, and checkpoints rewind any turn.

Which operating systems does Foreman run on?

Foreman runs on macOS (Apple Silicon, signed and notarized) and Windows. You need Git on your PATH and a login or API key for at least one agent.

Can Foreman run coding agents on a schedule?

Yes. Any saved prompt template can run hourly, daily, weekly or monthly in your time zone, and opt-in auto-fix sends failing CI logs to your agent. Scheduled runs use Claude Code and happen while Foreman is running, which can be without a window or from login.

Can I control Foreman from my phone?

Yes. Pair a phone with a QR code to answer approvals, steer agents, comment on diffs and read terminals, live. On the same Wi-Fi it connects directly, and anywhere else through your own Tailscale network. Native iOS and Android apps are coming soon.

Can coding agents run in a cloud sandbox like E2B?

Yes, as an experiment. A workspace's worktree and agent can run in a sandbox on your own E2B, Daytona, Vercel Sandbox or Docker account, for Claude Code, Codex and OpenCode. Provider keys stay on your computer and never go into the sandbox.

Which integrations does Foreman have?

GitHub (pull requests, merges, CI and reviews), GitLab, Linear, Slack and Langfuse, plus MCP. Each connection has a default per service, each repository can pick its own, and you choose what agents may do with it: nothing, read, or read and write.

How is Foreman different from Conductor?

Foreman is open source under Apache-2.0 and runs on Windows as well as macOS, while Conductor is a closed-source Mac app. Foreman also drives OpenCode and any ACP agent, and adds flows that loop until checks pass, AI employees and a scriptable command line.

How is Foreman different from T3 Code?

Both are open-source GUIs for coding agents. Foreman adds flows (agent graphs that loop until tests pass, with human gates), AI employees with a git-backed memory, a foreman CLI where every command takes --json, phone approvals and remote sandboxes.

Can I script Foreman or use it in CI?

Yes. The foreman command starts workspaces, sends messages, answers approvals, opens and merges pull requests and runs flows. Every command takes --json and exit codes are the same for every command.

Also hiring

Now hiring: humans

Foreman is 31 releases and three weeks old, built in the open under Apache-2.0. Star it so you hear about the next one, fork it, or pick up a good first issue. No interview.