Human review, only when it matters

Let builders build.
Let agents ship.

Forgecroft clears routine changes, stops dangerous ones, and sends only the changes that need judgment.
Works with any coding agent. If Forgecroft cannot tell, it asks for a human.

Request early access See the decision
payments-api · change #482 · adds webhook retry
Existing CI all greentests · lint · build
Forgecroft gate Human neededduplicate billing: retries re-emit queue events; invoice-worker never dedupes
The condition hold the retry flaguntil invoice-worker proves idempotency
orders-api · change #533 · drops unused column
Existing CI all greentests · lint · build
Forgecroft gate Blocked3 services still read it: ledger, reports, invoice-worker via views
To unblock migrate the readers firstthen the drop is safe to ship
billing-service · change #517 · edits payout schedule
Existing CI all greentests · lint · build
Forgecroft gate Human neededregulated payment path: payout logic sits in PCI-audited scope
The condition human approval requiredpolicy traced to payout-policy.yaml · line 42 · commit 8f3c2ae
web-console · change #541 · fixes typo in empty-state text
Existing CI all greentests · lint · build
Forgecroft gate Clearedcopy-only change: no logic, no data path, no control in scope
The outcome ready to mergeno human review needed; reason recorded

01How it works

Three outcomes. Only one interrupts a human.

A change is cleared, blocked, or sent to the right person with one reason attached.

Change
Opened by an agent or a person
Check
Independent of the author
Decide
Clear, review, or block
Review
Only when judgment is needed
Ship
Routine work keeps moving
Routine changes keep moving → ← Risk and uncertainty go to a person

02The decision

The answer at the merge

One result tells the workflow what to do next and tells a person why.

Clear

Can this merge?

Routine changes receive a clear path forward without waiting in a review queue.

Escalate

Does this need judgment?

Ambiguity, sensitive work, and missing context trigger review with the open question attached.

Block

Why did it stop?

Dangerous changes stop before merge with the violated requirement and a concrete next step.

03How it's different

Human attention becomes the exception

AI review helps people read changes faster. Forgecroft decides which changes still need them.

Agent workspaces

Help a person review faster

Review tools summarize changes, find issues, and suggest fixes. A person still makes the final call.

Forgecroft

Decide when a person is needed

Forgecroft gives the workflow one action: merge, retry, or escalate. Routine changes keep moving.

04Trust

The coding agent does not approve its own work

Forgecroft checks changes independently of the coding agent and keeps the reason with the decision.

Uncertainty triggers review

Missing context triggers review. Forgecroft never turns an unanswered question into approval.

Impact comes with the answer

Reviewers see what the change could affect without reconstructing the system themselves.

The reason is inspectable

Decisions point to the source behind them, so a person can verify why the change moved or stopped.

05Outcomes

Return attention to the work that needs it

The workflow stays technical. The result is legible across the company.

Builders

Spend review time on design and judgment instead of routine agent output.

Engineering leaders

Turn faster code generation into merged work without growing the review queue at the same rate.

Security and reliability

Keep an independent reason for autonomous approvals, blocks, and human escalations.

Find out what can merge without you

Tell us about your repositories and current review queue.

Request early access