Six repositories, one answer.
Limn lights up what a fortnight of work across every repository you own actually came to — where it happened, how long it waited, and who was holding it — and then what the review bots layered on top cost and returned. It shows what the data can carry and refuses the rest, which is most of what this category sells.
Sections 02 to 05 are free on every tier. Run the whole free product on your own machine with npx pierre-review.
Five questions, and one place that answers them.
Where is the effort going, what is stuck, how long are people waiting. Answering those across a set of repositories means counting the same things the same way in each of them — and then separating what people did from what automation did, because agents now author a real share of the pull requests and several vendors comment on every one.
The questions beside this are the ones actually asked in a staff meeting. Limn answers each with a stated population and a stated window, or says why it cannot.
- Where is the work actually happening across the repositories I own?
- What is stuck right now, and is it stuck on a reviewer or on the author?
- How much of the review load is people, and how much is automation?
- Which of the bots we pay for is earning its seat?
- What do I put in front of a director on Friday without writing it by hand?
Never · a developer leaderboard. There is no such screen in the product.
The state of play, at no charge.
Throughput, lead time, time to a first human review and merge-time build health for the whole workspace — with people counted apart from automation, because a blended median describes no pull request anyone wrote.
Time to a first human review is the number most worth holding. It is the part of the wait that is entirely yours to spend, it sits ahead of everything else in the queue, and no bot approval is allowed to count towards it.
A repository added part-way through a window is marked rather than pro-rated, and where a pull request’s size was never observed it says “size unknown” rather than counting it as zero.
Free, on every tier · this whole section, and the two that follow it.
Which repository is carrying the fortnight.
Two columns, two scales, two origins. A bar’s length is a ratio within its own column and never a number you can read across — the two measures are counted in different units, and putting them on one axis would invent a comparison.
Pull requests opened splits people from automation in the bar itself, so you can see which repositories the agents are working in without doing arithmetic.
The repository name is written out in full and wraps. The rotated, truncated axis label it replaced rendered six of seven real repositories as “…tric-backend”.
And how much of it could break something.
The open pull requests in each repository, split by how far a change could reach — worked out from what the diff touches: migrations, schemas, interface definitions, infrastructure, and whether files that usually move together are missing from it.
It is a snapshot of right now rather than a window, and it says so. Drafts are included, and the draft count is disclosed, because the tile above it excludes them and two numbers that disagree need to explain themselves.
Two repositories report no reading rather than a zero. The co-change index only exists where the history supports one, and a zero there would read as a clean bill of health.
Find every bot. Name it. Price it. Free.
Limn detects every automated account touching your repositories and separates the review vendors from the dependency bumps, the quality gates, the release automation and the agents that write code — and lets you correct any of it by hand, with your judgement winning in both directions.
That split is what makes the human numbers above trustworthy. A merged dependency bump and a merged feature are both “a merge” until something tells them apart, and on real data the bumps alone dragged a reported median pull-request size from 142 lines to 68.
Type in what each one costs a month and the figure is stored per workspace. The grade on every bot comment is free too — a purpose-trained ModernBERT classifier, served as int8 ONNX on ordinary CPU, which — graded blind against a human expert — picked the same severity seven times in ten, where the vendor’s own badge managed fewer than five. The ruler is never the paid part.
A report per sprint, that you can forward.
A stored artifact for each completed sprint or calendar month — fifteen metrics, a like-for-like comparison against the previous period, the biggest movers, and a short written summary. Every event in it is timestamped inside the window at both ends, so the document stays reproducible after you send it.
The comparison is recomputed over the repositories tracked in both periods and says so on the row. Otherwise the delta would be measuring which repositories you onboarded rather than what the team did — and the headline figure and the change would be two different populations wearing one number.
The forecast refuses at calendar-month grain outright, because months run 28 to 31 days and a trend line fitted through them reads February as a dip every year.
Where the waiting happens — without naming anyone.
Every hour a pull request was open belongs to somebody: a reviewer who has not looked, an author who owes a response, or nobody — approved and waiting to land. Chronology attributes all of them and reports the split per repository.
A bot action never moves the ball, and pull requests no person ever touched are excluded and reported separately — on real data that is 46% of merges, and folding them in would report a reviewer problem that is an automation fact. It speaks up only when a court is both lopsided and slow, because a share without a magnitude invents a crisis in a healthy repository.
No person is named anywhere in it, and the server sends no actor ids at all — that is structural, not a setting. The People report, for one-to-ones, is the opposite by design: you pick who, the sections are alphabetical, and there is no ranking and no comparison table in it.
Sixty-six merged pull requests a person actually worked on. The never-human-touched population is reported on its own, never blended in.
What the bots cost, and what they returned.
A keep, tune or noisy verdict per vendor, built from how much of what it said your team acted on — and dollars per acted-on comment, divided by one stated calendar month at both ends, with both halves of the division printed on the card.
Beside it, over-calling as a weekly line: how often a vendor badged something major that our model read as minor. A bot that badges nothing is named and omitted rather than drawn as a zero.
It refuses where it should. A repository younger than the window, a reviewer with nothing acted on, or fewer than ten acted-on comments each withhold the per-comment figure on their own, while the honest counts still render.
The receipt · 488 bot threads in a fortnight, 62% acted on, six vendors, one of them worth dropping.
And how that compares to everyone else.
Your vendors placed against a cohort of 2,204 repositories, banded by how busy a repository is, so a quiet service is not compared against a monorepo. Every percentile carries the number of repositories behind it and which band it sat in.
Where the cohort is too thin to answer, it says so in words — there are fourteen distinct refusal sentences in this panel, and they are the feature. A percentile drawn from four repositories is a number that will move on its own next month.
Free answers one pull request. Pro answers the fortnight.
Limn is open core. The free tier is the product, not a trial — it has no repository limit, no user limit and no expiry, and every action you take on a pull request lives in it. Pro is what happens when you need the same truth one grain up: across repositories, across people, across a period.
What is waiting on me?
The whole worklist, ranked. My Turn, review requests, red builds, conflicts — across every repository.
Same list, with a written reason on each of the top items and a line saying what can safely wait.
What happened on this pull request?
Every thread triaged — untouched, replied, likely addressed, resolved. Blast radius, blockers, merge verdict.
A short written summary, a read on what a red build means, and a check on whether a comment was really addressed.
Can I act without leaving?
Reply, resolve, approve, merge, merge-when-ready, update from trunk, resolve conflicts hunk by hunk.
Everything free does. Acting on a pull request is never the paid part.
Which review bots are worth their seat?
Detect every bot, name it, set its role and record what it costs — and an independent severity grade on every comment it leaves.
The per-bot scoreboard: keep / tune / noisy, dollars per comment your team acted on, over-calling as a weekly line.
How does that compare to other teams?
—
Your bots placed against a 2,204-repository cohort, with a written refusal wherever the cohort is too thin to answer.
How is the team doing?
Flow metrics for the workspace, activity and reach by repository, open pull requests and trunk health.
A stored report per sprint or month, compared like-for-like against the last one, with the coverage stated.
Is a repository about to bite us?
Default-branch health across every repository, the commit that broke it, and the pull request that landed it.
Everything free does. Trunk health is never the paid part — it is the thing you need at the worst moment.
Where is the time actually going?
—
Chronology: every open hour attributed to a reviewer, an author, or waiting to land. Nobody is named.
What do I take into a one-to-one?
—
A report with a section per person you pick. Preparation, not a ranking — there is no leaderboard anywhere in it.
Can I keep the noise down?
Bots hidden by default, notifications only for work that is personally yours, and a per-repository mute.
Everything free does, plus a Slack digest per workspace if you would rather read it there.
Where does it run?
npx pierre-review on your machine, or the hosted service. No repository limit, no user limit.
The same two places. Pro unlocks on whichever one you already use.
Pro lists at $25 per user, and is free for a month to anyone who asks for it. A user is someone who signs in — bots never count, and there is no per-repository charge.
What a free account actually sees. The paid panes are listed and locked rather than hidden, so you can tell what you are not buying before you buy it — and the free metrics above them are the ones that moved here off another tab.
Start free. Every repository, nothing missed.
Sign in with GitHub and the first repositories are on screen in a couple of minutes, or run the whole free tier on your own machine and keep the data there. For Pro, send a message — it is free for a month, for asking.
Card payments are not switched on yet, so Pro is not sold from this page — it is given, a month at a time, to anyone who asks. No card, nothing to cancel. Or run the free tier locally: npx pierre-review