Think inside your AI world.
CircleCI through Unl
CircleCI shows the failed job and the logs behind it. Whether that failure is a known flake or a real block is your call - and 'it's red' isn't the same as 'it blocks the merge'.
CircleCI through Unl tests failure logs and flake history against the merge rule the tech lead has ratified, so 'is this a flake or a block?' comes back verdict-shaped, naming what hasn't cleared.
What CircleCI holds
What CircleCI through Unl retrieves, unedited:
- Latest pipeline status for a project
- Build failure logs, job by job, step by step
- Job test results
- Flaky-test history across recent runs
What the naked read gives you
The CircleCI MCP server gets latest pipeline status, pulls structured build failure logs down to the command and exit code, retrieves job test results, and can flag tests with a history of flaking. That's the real CI record, not a guess at it. What it doesn't resolve is whether this particular red build should stop a merge: a test with three flaky runs last month might still be the one that's genuinely broken this time, and CircleCI's history is a pattern, not a ruling. The build is red. Whether it blocks is a call the pipeline data alone doesn't make.
The frame judges the data it is given; it does not verify the source’s accuracy.
What changes when CircleCI is read measured
The tech lead ratified in Unl that a failing test only counts as a merge-blocking break when it hasn't flaked in the last ten runs and the failure isn't isolated to a known-unreliable suite - because flaky tests were blocking merges the team had learned to ignore.
'Is this a flake or a block?' returns, measured: a block - this test has passed all ten of its last ten runs, so today's failure isn't the known flake pattern. CircleCI supplied the failure log and the flake history; your ratified rule supplied the ten-run rule that separates flake from break.
And back again
The flake-or-block verdict writes back into Unl with the test name, the run history it was checked against, and the outcome. The next time this suite goes red, the tech lead isn't re-litigating from the raw logs - the ten-run rule and its last verdict are already in Unl, reusable or adjustable as the suite's real reliability changes.
The answer comes back measured against what you already decided, and why.
The lane is live and open to this tool today: one box, paste anything. If it speaks MCP, Unl can reach it. Readings arrive unprompted, the data beside the criterion; Unl is a courier, not a warehouse, and keeps only your keys and the frame.
Questions people ask
Can AI tell me if a failing test blocks the merge?
Not from the log alone. CircleCI through Unl reads the failure log and the flake history; whether it blocks depends on the run-history rule you ratified in Unl. A test with ten clean runs and one with three recent flakes return different verdicts on the same red build.
How do I connect CircleCI to Claude?
CircleCI publishes its own MCP server, documented at circleci.com/docs/guides/toolkit/using-the-circleci-mcp-server, installed with a personal API token. Add Unl alongside it, since the flake-versus-break rule the failure gets checked against isn't something CircleCI itself stores.
Does Unl rerun the pipeline or merge the branch?
No. Unl pulls pipeline status, failure logs and flake history from CircleCI and stops there - it doesn't trigger, rerun or merge anything. Whatever happens to the build after the verdict is a separate, human decision.
What this is
Think inside your AI world — you stay in command
Unlimitless (Unl to friends) holds what you've settled, reads what your tools are showing, and catches what's changed out in the world — and hands your AI whatever bears on the work, the moment it's needed, without you asking. The right thing, in front of the model, unprompted, with you in command of the call. So you keep moving toward what you set out to build, on top of everything you've already decided.
It plugs into Claude, Claude Code, ChatGPT and Cursor as an MCP connector. Quick to connect, in a couple of steps.
Unlimitless is open now to invited Alpha. Apply for the Beta waitlist to come in ahead of the full launch:
Alpha is invite-only · free at launch.