Product
A second reviewer that shows its working.
Forseti reads every pull request at the depth you set, tells you the price before it starts, and posts only findings that survive a second look. Here is what happens, step by step.
- 1WebhookSignature verified, event recorded, queued in under a second.
- 2EstimateSize × depth = a price and a time, checked against your caps.
- 3TriageA fast triage pass sorts files by risk; lockfiles and generated code skip ahead.
- 4FindA deep reasoning pass reads risky hunks in context; Deep adds four specialists.
- 5VerifyEach finding must cite file:line, a failure and a fix, or it goes.
- 6PostOne review, a living summary, one check run for the head commit.
- 7LedgerFinding IDs and states carry across pushes; spend is settled.
Every capability
Built for the whole path to merge.
Depth you choose, a price you see first, findings that survive a second look, and the context around the PR: the issue it closes, the dependencies it adds, the people who should look.
Review
What it reads, and how hard.
- Reviews & depthChoose how hard every PR is read: Glance, Standard or Deep.
- Attention map & reviewersThe hunks that need human eyes, and who should look.
- Domain packsMoney, migrations, tenancy, PII and architecture checks.
- Intent checkIn testingDoes the PR do what its issue asked? Evidence per criterion.
- Supply chainIn testingNew dependencies, advisories, licences and typosquats.
Control & trust
What it costs, and what it remembers.
- Price GuardThe price on the PR before anything is spent.
- Review LedgerOne record per PR: every finding keeps its ID across pushes.
- Team memory & precisionIn testingIt learns what your team dismisses, and shows its hit rate.
Workflow
Where it meets the rest of your day.
Depth
Four depths, each with a known cost.
Prices are core's estimate for a medium PR of about 1,000 changed lines. Smaller PRs cost less; every PR is quoted on its own size.
| Depth | What runs | Medium PR | Typical time |
|---|---|---|---|
| GlanceLive | A fast summary of the change with a self-check. No line comments. | £0.03 | about 38 seconds |
| StandardLive | A fast triage pass, a deep reasoning pass, then a second, independent check of each finding. Low-risk files get the fast pass. | £0.79 | about 1.6 minutes |
| DeepLive | Standard, plus four specialist passes and an adjudication step. Static analysis arriving next. | £4.64 | about 2.5 minutes |
| MolecularEarly access | Early access. Deep, plus an agent in a sandbox that runs the related tests and attaches a failing test to each serious finding. | £10.23 | about 9.0 minutes |
Recommended pricing; final numbers follow measured runs on a golden set of real PRs. About Molecular early access.
Where it shows up
Zero clicks on GitHub. Full control in the dashboard.
Developers never leave the pull request. Owners, admins and billing get one place for policy, spend and quality.
On GitHub
- Check run
- One per head commit, updated in place: the estimate first, then the verdict.
- Inline comments
- A few strong findings with one-click suggestions, within your noise budget.
- Living summary
- What changed, what it found, intent, attention map and suggested reviewers.
- Commands
- @forseti review deep, confirm, pause, resume, config, explain, ask, split, release-notes.
In the dashboard
- Reviews & PR detail
- Every run, its cost and findings, with the diff and the Ask drawer.
- Policies & repos
- Depth by repo, branch and path; packs; strictness; dimensions.
- Spend
- Ledger, caps, ceilings, forecast and top-ups.
- Insights & Quality
- Cost per bug caught, precision trend, learnings and noise budget.
- Readiness
- Conflicts, CI, approvals and Forseti's verdict for every open PR.
- Team & settings
- Roles, invites, integrations, retention and the audit log.
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.