Foreman
macOS · Windows · Linux · v1.5.3 · free

You stopped writing code. Your tools didn’t notice.

You can start five agents. You can’t watch five. Foreman is the control room — every agent, both CLIs, every machine you own, in one window.

macOS, Windows and Linux — all formats.

guest-checkout$16.50
↳ storefront-web · ▣ studio · Running
Add guest checkout — email only, no account creation.
Claude Codeclaude-sonnet-4-5
Read the checkout flow — the account step is the only blocker.
Editsrc/checkout/steps/GuestEmail.tsx+42 −6
Plan ready for reviewNeeds approval
4 steps · touches 6 files — add an email-only guest path, keep account creation as a post-purchase offer, cover both in the checkout tests.
Request changesApprove & build
✦ Skills ▾Ask every time ▾Sonnet ▾Medium ▾Directives ▾🎙
Ship it — and start on the webhook retries next.
⌘↵Send ↑

The loop: you direct, agents build, nothing merges without you. Watch the real app

The full 80-second walkthrough.

The premise

You haven’t opened a file in weeks

Editors were built for people who type. You review, you decide, you redirect. Foreman is built for the job you actually do now: diffs to read, plans to approve, questions to answer, cost to watch — and no blinking cursor anywhere.

Orchestration canvas

Your whole operation, on one canvas

The default view is a map, not a list: every project a box, every session a card with live status, git state, and cost. The difference between running agents and overseeing them.

✦ MemoryTriage2Planning1Building3
7 memories
storefront-web4+
2 queued
checkout-retries
studio$8.18
checkout · build
studio$2.40
checkout-retriesracing · 2 lanesCompare
lane-a
studio$0.91
lane-b
studio$1.36
Parallel fan-out

Race the same task, keep the best attempt

Fan a hard task out to four agents in parallel, each in its own isolated git worktree. Compare the finished diffs side by side, keep the winner, discard the rest in one click.

storefront-web3+
refund-webhookracing · 3 lanesCompare
refund · L1
studio · opus$0.91
refund · L2
studio · sonnet$0.42
refund · L3
studio · haiku$0.11
Project review

Audit your whole codebase with one click

Fan out a fleet of review agents — dependency health, test gaps, architecture, security, dead code — each measuring with your repo’s own tooling. The result is one structured report where every number cites its evidence, and every finding has a fix button.

Findings
10
Critical + high
6
Run cost
$1.67
Duration
3m 19s
Dependency health5
Test coverage & gaps5
critical
electron is 9 majors behind with live advisories
pnpm audit lists ~30 advisories against the installed range, several high.
package.json:70pnpm audit
high confidence
▸ Start session to fixCopy as promptEst. ~2 days
Scheduled tasks

Put your agents on a schedule

Hand an agent a standing job and a time — once, every morning, or every Monday at 09:00 — and Foreman runs it unattended, leaving the result waiting for your review.

Scheduled tasks

Run an agent prompt on a cadence. A missed run fires the next time Foreman opens.

+ New schedule
Weekly CLI release scan⑂ Worktree
Every Monday · 09:00· next in 3d
🗑

What's new in the Claude CLI this week, and should we adopt it? Summarise the changelog and open a PR if a flag we use changed.

Completed· changes ready· 2d agoEnabled
Nightly repo health check
Every day · 02:00· next in 9h
🗑

Run the test suite, check for outdated deps and flaky specs, and leave a short report in the session.

Running…· started 4m agoEnabled
Permissions, plans & questions

Approve the work, not just the diff

Agents pause and ask in the chat, not in a buried terminal prompt. Approve each tool call, review the plan before a single file changes, answer questions with one click.

›_Bashpnpm db:migrate --env stagingNeeds approval
DenyAlways allowAllow
?Claude has a question for youNeeds answer
Auth
Which auth strategy should I use?
JWT + refresh tokensSession cookiesSomething else…
Or type your own answer…
SkipAnswer ↑
Plan ready for reviewNeeds approval
4 steps · touches 6 files — migrate the retry queue to exponential backoff, cap at five attempts, and add a dead-letter table.
Request changesApprove & build
Project memory

Agents that remember your project

Curated memory per project — conventions, decisions, gotchas — that you pick and attach when a session starts. Yours to browse and edit; nothing gets injected behind your back.

Load context

HandoffsMemories
storefront-web
API conventions
Route handlers live in src/api; zod-validate every body.
Deploy checklist — staging
Run migrations before flipping the feature flag.
Postgres gotchas
The orders table trigger fires on soft-delete too.
Add to context (2)
Git, built in

Review the diff, send it back to the agent

Read the diff of every change line by line, comment on any line, and send it straight back to the agent to fix. Plus the full git muscle — merge, rebase, cherry-pick, conflicts.

Opus 4.8 ▾High ▾3 of 5 files✦ AI review2 comments added
src/checkout/GuestForm.tsx
@@ -18,7 +18,10 @@ export function GuestForm() {
2020 const submit = (e) => {
2121 e.preventDefault();
22 send(email);
+22+ if (!isEmail(email)) return setError(email);
✦ AI review · line 22 — setError is called with the raw email instead of a message — the invalid-email branch renders the address back to the user as its own error text.
23+ send(email.trim());
2324 };
webhooks.ts:64–68The retry loop swallows the 5xx body, so a permanent failure is indistinguishable from a throttle here.
Note to the agent (optional) — how to treat these comments
Send 2 comments to agent ↑
Session from phone

Steer your agents from your phone

Share a running session and open it in any browser — read the live transcript and send the next turn while work keeps running on your Mac. End-to-end encrypted, nothing stored.

End-to-end encryptedNothing stored on our serversPure pass-throughRevoke anytime

Share Laptop

Make this machine available to your other devices. They connect through your Foreman account — encrypted end to end, with nothing exposed on your network.

Sharing
studio — this Mac
On
Connected — 1 device driving this machine.
Sessions keep running here even after it disconnects.
This laptop’s safety code
4f2c 91ab 77de 05b3 · 8c14 6ea0 d92f 31cc
guest-checkoutlive
Claude Code
Tests pass — 24 green. Ready for your review.
Ship it — and start on the webhook retries next.
Reply…

Also in the cockpit

See it run, right there

A real terminal per project, and one-click preview of running dev servers.

From GitHub issue to working session

GitHub issues to briefed sessions — own branch, PR that closes the issue.

Every machine you own, one window

Every Mac and build server you own, on one canvas.

Know what every agent costs

Live per-session and per-project spend, with a daily heatmap.

Not married to one agent

Claude Code and Codex, managed from the UI — pick per session.

Subscription or API key — you pick, per session

Subscription for interactive work, API key for automation — per session.

OS notifications

Get pinged when an agent needs you or finishes — even with Foreman in the background.

Session branching & handoff

Fork a session to explore, or hand off a curated context digest to a fresh one.

Agent browser

Give a session its own browser — watch it work, and take the wheel when it hits a login.

Directives

Reusable standing instructions attached per session — your rules on every turn.

Voice dictation

Talk to your agents — on-device Whisper dictation straight into the composer.

MCP servers & plugins

Every MCP server visible in one place; log in where auth is needed, no config spelunking.

Setup wizard & diagnostics

CLIs, logins, and git checked and fixed from the UI — no config-file spelunking.

In-transcript search

Search inside any session’s transcript and jump straight to the match.

Streaming chat

Watch agents think and act live, with persistent history that survives restarts.

See all of it in detail →

Security posture

Your code never meets our cloud

Local-only by design

Foreman talks to your machines, not to our servers. Code, prompts, and transcripts never leave your network — cost figures only if you switch on cross-device tracking.

Keys stay in your Keychain

Claude tokens and SSH credentials live in the macOS Keychain, encrypted at rest. Pairing codes are single-use and expire.

Bring your own key

Foreman drives the Claude Code you already pay for — your Anthropic account, your data agreements, your rate limits.

End-to-end encrypted relay

The phone and browser path is sealed on your devices. The relay routes ciphertext it cannot read, and stores nothing — your Mac stays the only source of truth.

Where Foreman stands

The parts nobody else ships together

vs Claude Code Desktop

  • It runs Claude. Foreman runs Claude Code and Codex side by side — one permission model, one memory, one cost view across both, and you route each task to whichever agent is better at it.
  • Your other Macs and build servers join the same canvas. Anthropic’s app only ever sees the machine it’s installed on.

vs the Codex app

  • The same trade in reverse: the Codex app is locked to OpenAI, so a second vendor means a second window and a second set of habits.
  • Foreman shows what every agent actually costs, across both vendors and every machine — one number, not two dashboards that don’t add up.

vs Conductor

  • Phone steering, scheduled tasks, and multi-machine all ship in Foreman today.
  • On top: issue intake, parallel fan-out with compare-&-keep, one-click multi-lens project review, and per-session subscription-vs-API-key billing.

Accurate as of August 2026 — check each product’s site for its current state.

Pricing

Foreman is BYOK — you already pay Anthropic for Claude Code. Foreman charges for orchestration, not tokens.

Free

One machine, no account

$0

  • Unlimited local sessions, 3 running at once
  • Streaming chat & diff review
  • Terminal, preview, one agent browser
  • Cost tracking

macOS, Windows and Linux — all formats.

Pro

Every machine you own

$19 / month

  • Everything in Free
  • Unlimited sessions running at once
  • Unlimited agent browsers, one per session
  • Multi-machine fleet (SSH + LAN)
  • Live fleet graph
  • Session from phone & browser
  • Fleet-wide cost analytics

Billed monthly in USD, cancel any time. Model usage is not included — you pay Anthropic or OpenAI directly for the agents you run. Full pricing details.
Buying for a team? Seats and a single invoice are coming — your sessions and prompts stay yours, only aggregate spend is ever shared. Tell us what you need.

Foreman — the control room for Claude Code and Codex agents