Skip to content

Price GuardLive

The price is on the PR before anything is spent.

Every review is estimated from its real size and depth, quoted on the PR, and held against your caps before a model is called. A cap never stops a review in silence: it drops to Glance and says so.

Connect GitHubRead the docs
Check run · Forseti / review
Example

Standard review · 640 lines · 9 files

estimate £0.53 · reserved £0.66 (p90) · about 01:21

Quoted before it runs
This PR£0.53 / £5.00 · per-PR cap
@maya this month£31.20 / £40.00 · developer ceiling
payments-api this month£118.00 / £150.00 · repo ceiling
i

Over the org month cap: reviewed at Glance instead. Comment @forseti confirm to run Standard for £0.53.

Dismissed as incorrect: “Unused import”+4 credits back

How it works

Step by step.

The price on the PR before anything is spent.

  1. 01EstimateReviewable lines and files, depth and dimensions give a p50 and p90 price and a time. The p90 is reserved.
  2. 02Check capsPer-PR, per-repo-month and per-organisation-month caps, plus ceilings per developer, team and repo.
  3. 03Degrade, don't skipOver a cap, the review runs at Glance with a notice, or pauses with one neutral check. An explicit Deep above the cap asks for @forseti confirm.
  4. 04SettleYou pay the measured cost, never more than the reserve. Re-reviews charge only for new changes.

In detail

What it does, exactly.

  • No auto-refill. Credits are 1p each, and you top up when you choose.
  • The reservation is the most a review can charge, shown before it starts.
  • A wrong finding costs you nothing: it is credited back automatically.
On GitHub
The estimate in the check run, cap notices, @forseti confirm
In the dashboard
Spend, caps, ceilings, forecast, ledger

Forecast

Next month's spend forecast from the last 30 to 90 days of runs, per repo, on the Spend page.

Ceilings per developer, team and repo

Set a monthly ceiling for anyone or anything. On a hit, reviews degrade to Glance with a notice on the PR.

Credit-back for wrong findings

Dismiss a finding as incorrect and its share of the run's cost comes back as credit: once per finding, capped at 20% of monthly spend.

Cost per bug caught

Insights divides spend by the blocking and should-fix findings your team accepted or fixed.

Works with

Next: Review Ledger
Reviews & depthChoose how hard every PR is read: Glance, Standard or Deep.Learn moreTeam memory & precisionIt learns what your team dismisses, and shows its hit rate.Learn moreReview LedgerOne record per PR: every finding keeps its ID across pushes.Learn more

Put Forseti on your next pull request.

Connect GitHub in four clicks. Your first review starts by itself on your latest small open PR, priced before it runs.

Connect GitHubBook a demo