A ticket goes in. A verified pull request comes out. A person approves every step.

Harborline runs a storefront. This is its engineering cockpit: an AI pipeline that researches each ticket, writes the fix, proves it with the service's own tests and opens the pull request — stopping at a human gate each time.

Enter the demo → Take the guided tour

Guest access, read-only. No sign-up, nothing to install. Harborline, its people and its tickets are fictional; the code and the runs are real.

HBL-101 · harborline-ticket-to-fixcartservice

Research the ticketapproved
Implement the fixapproved
Verify the fixPASS
Push and open the pull requestPR #1
Final reviewawaiting approval

Grounded in the code

The agent reads the services the ticket points at before it writes a line. Research names files, reproduces the bug, lists acceptance criteria.

Proven before it ships

Every fix is built the way its Dockerfile builds it and run against the service's own tests. A failing verdict sends the run back, not forward.

A human at every gate

Approve, or send it back with a note. Nothing is pushed, merged or posted without a decision — and every decision is on record.

Any stack, any tracker

Harborline's platform is Go, Java, C#, Python and Node. The same pipeline works on GitLab or GitHub, Jira or another tracker.

What the tour shows you

The runs listEvery ticket in work, which flow handles it, how far it got and what it cost.
Research, grounded in the codeThe document an analyst approves before any code is written.
The fix as a diffThe change in the cart service, with its unit test, and the full transcript of what the agent did.
VerificationBuild, tests, the ticket's scenario reproduced before and after — the verdict QA signs off.
The pull requestPushed and opened on GitHub with the evidence in its description; the ticket hears back.
Tools and metricsAsk about the codebase, review a PR; then the numbers the runs add up to.

Enter the demo →