Open-core and free. Pro from fifteen dollars a seat.
The dashboard is free, local-first, and yours forever — and the bot receipt is free with a hosted sign-in. Pro reads the board for you; Pro+ closes the loop with Claude. Each costs less than the review-bot seat it audits.
Free
Open coreEverything you need to see your team — and your bots — clearly.
- Unlimited repos & contributors
- Activity console + cross-repo feed — the view Limn opens on
- The timeline board — repos down the side, time across the top
- The bot receipt — severity, category, cost & overlap from an independent ML model (hosted today; coming to the local install)
- Review-thread states — resolved / likely addressed / replied / untouched
- PR detail with inline diffs, CI logs and suggested reviewers
- Reply, resolve, approve, request reviewers, rebase & merge from Limn
- Cross-repo search, workspaces & comparison metrics
- Runs 100% local — npx pierre-review, your data never leaves your machine
Pro
The intelligence layerEverything in Free, read for you: judgement on top of the receipt.
- Attention & risk digests — what the bots posted, what was acted on, which threads need a human
- Thread validity checks, inline while you review
- “Was this addressed?” — immediate, with a confidence gauge
- Themes & reports across human and bot reviews, by severity
- Chat with your repos + ad-hoc charts — pin the keepers, re-run past reports
- CI failures summarised to root cause, inline
- Reviewer suggestions from commit history — requested in one click
- Flow metrics with drill-downs · Slack digests · Jira & Linear links
Billing via Stripe · cancel any time · a declined card downgrades you gracefully to Free — nothing is deleted.
Pro+
The full loopEverything in Pro, plus the agentic loop — review, fix, push, one app.
- Everything in Pro
- Claude Review — context-aware, deep or quick, routed automatically
- Review memory — every run learns from what you kept, cut and reworded
- Reword or simplify any finding — post one review, in your voice
- AI Fix — pick the comments that matter, review the diff, push
- CI auto-fix — diagnosis to pushed fix in two clicks
- Runs on your Anthropic key — your models, your spend, metered in-app
BYO Anthropic key — the model spend is yours, at cost, metered transparently in-app. No surprise bills.
A seat is someone who signs in — watchers and leads count, bots never do. Reassign freely.
Pro+ runs on your Claude. That’s the point.
Claude Review and AI Fix spend real model tokens on your code. On your own key that spend is yours at API list price, metered in-app — Limn never marks it up, never caps a heavy week, and never has your code transiting anyone’s account but yours. Claude Sonnet 5 by default; Opus 4.8 for the hardest runs; Haiku 4.5 for quick passes.
How the agentic loop works →Review bots run $24–30 a seat and add comments. Limn adds signal.
We’re not selling another reviewer. Pro is half the price of the bot seat it audits — and the receipt tends to pay for itself the first time you tune, or cancel, a noisy bot.
The questions worth asking before paying anyone.
Is free really free?
Yes. Limn is open-core: the free dashboard is the product, not a trial. The Activity feed, the timeline, PR detail with real GitHub write actions, the bot receipt — all of it stays free, with no repo or seat limits. The paid tiers add layers on top.
Why is the bot grading free?
Because a measurement you have to pay for is a measurement you'd doubt. The grading model is Limn's own and runs on plain CPU — no LLM meter — so it costs next to nothing to serve, and the paid tiers are the layers on top of the ruler, never the ruler itself. It runs in the hosted service today; grading in the local install is on the roadmap.
What counts as a seat?
Anyone who signs in. The lead who only watches counts; the bot never does. Seats are reassignable, so a departure doesn’t strand one.
Do I need an Anthropic key?
Only for Pro+ — Claude Review and AI Fix run on your own Anthropic API key, with usage metered transparently in-app. Everything in Pro (digests, validity, themes, chat, CI diagnosis) is included in the subscription, no key needed.
Do you rank my engineers?
No. Everything is team- and process-grain: bots get graded, people don’t. Mirrors, not scorecards — by design, not as a setting.
What happens if my payment fails?
You drop to Free, gracefully. Nothing is deleted, nothing locks you out of your data — you keep the whole dashboard, receipt included, and lose only the paid layers until billing is sorted.
Do you train on my code?
Never. Your data is yours — private and confidential, forever. Limn never trains on it, never shares it, and in local mode it never even leaves your machine. The grading model was trained on public GitHub bot reviews, not on you.
Can I self-host?
The Open Core tier, fully — npx pierre-review runs the whole free dashboard on your machine, and the multi-tenant cloud deployment is self-hostable too. Pro and Pro+ are subscriptions that unlock on any deployment as the hosted rollout lands.
Fifteen dollars a seat. Every repo. Nothing missed.
Start free and read this month’s receipt — upgrade when you want the board read, and the loop closed, for you.