Review pipeline results

What this means

The pipeline detail page is where you watch a pipeline run and review its result. Every pipeline has the same shape: a header card with the headline facts, then a row of tabs covering the timeline, AI calls, CI, clarifications, children, learnings, the diff, and diagnostics.

You don't need to read every tab on every pipeline. Most users only open Timeline and the merge request link. The other tabs are there when something needs explaining.

When to use it

• A pipeline just finished and you want to read the result.
• A pipeline is in Awaiting clarification or Awaiting approval and you need to act.
• A pipeline failed and you want to know why.
• You want to know how much it cost or how long it took.

Header

The card at the top of the page shows the most important facts. Bilbis only renders fields that have a value, so a fresh pipeline shows fewer fields than a finished one.

The right side of the header shows action buttons:

Tabs

Tabs appear in this order. Some only render when there's something to show.

Timeline

The default tab. A chronological list of state transitions and agent actions. Each row tells you which agent moved the pipeline and when.

Use this tab to:

• Watch a pipeline run live.
• Skim what happened, in order, after the fact.
• Find the moment a pipeline turned red.

If a state transition has an agent question, the question appears inline with an amber callout.

LLM calls

Every prompt and response the agents made. Each row shows the agent, the model, the cost, and how many tokens were used. Open a row to see the prompt and the response.

Use this tab when:

• A pipeline cost more than expected and you want to see why.
• An agent produced a wrong answer and you want to read its reasoning.

CI runs

The CI checks the pipeline triggered. Each row shows the provider, status, and a link out to the run. If a CI run failed, the failed jobs and their logs appear here so the agent (and you) can see the same thing.

Use this tab when:

• The pipeline is in CI fixing and you want to see what's failing.
• The MR has a red CI check and you want the actual error.

Clarifications

Only shown when the agents asked a question. The tab dot pulses amber while a question is pending.

Use this tab to:

• Read the question the agents asked.
• Reply, so the pipeline can resume.
• See the history of past questions and answers.

Some clarifications come with quick-reply buttons (yes/no, branch picker). Others are free text.

Children

Only shown when this pipeline spawned other pipelines. This happens when Let Bilbis pick routes a single task across multiple repos and the Repo Router decides to split it. Each child runs on its own and reports back.

Learnings

What the Review Agent extracted from this run that it thinks will help future pipelines: conventions, patterns, CI fixes, review fixes, and gotchas. The full list of learnings across all repos lives at Insights → Learnings.

Diff

The actual code changes the pipeline produced. Loaded only when you open the tab. If the pipeline opened an MR or PR, the diff comes from there; otherwise it comes from the working branch.

Diagnostics

Internal state Bilbis collected while running. Most users only need this tab when something failed and the failure reason isn't clear from the timeline.

Use this tab when:

• A pipeline failed and you want the underlying error.
• A pipeline is stuck and you want to know what it's waiting on.

Tab title

When a pipeline is running in a backgrounded browser tab, Bilbis updates the tab title with a glyph so you can tell what's happening without switching tabs:

Problems and fixes

Related pages

• Pipeline states - every badge explained.
• Create a pipeline - what each form field controls.
• Troubleshooting - recovery for stuck or failed runs.