FREE · CI HEALTH REPORT

How often is your CI actually green?

Get an audit of your real CI reliability. Reviewed by experts and benchmarked against the fastest-shipping teams, for free.

Read-only metadata access, authorized by a GitHub admin. See exactly what you grant →

MAIN PASS RATE · WHERE YOU STAND
Peer teams fastest shippers94.0%
You sample82.0%
Median industry reports70.8%

How often main is green is how often you can actually ship. Numbers above are illustrative.

The CI reliability platform trusted by teams that ship fast
Brex
Zillow
Faire
Cockroach Labs
Klaviyo
Retool
Google X
Zipline
THE METRICS THAT MATTER

Insights you can action on

No vague “developer productivity” but real numbers and benchmarks you can take to your leadership and inform your next steps.

PR
90.5%
MQ
93.0%
MAIN
87.8%
IND. MED.
70.8%

Is main shippable right now?

How often CI is actually green - across PRs, the merge queue, and main - measured against the industry median.

BetterUp went from 70% → 90%+ reliability
1.2 days

lost to 55 re-runs across 48 PRs

What do flaky re-runs cost you?

Flakiness makes you pay that CI wall-clock again and again, re-running green code. The real cost is engineers' time, not the compute bill.

Faire reclaimed 330 hrs in 35 days
Job #1
Job #2
Job #3
Job #4
Job #5
5-min gold standard · gray = waiting to start · long pole gates the PR at ~17m

How long does every change sit in CI?

Every PR waits on its slowest job - the long pole holds up everything behind it. Pure wall-clock, before a single re-run.

Caseware cut merge time 75% - 6+ hrs to 90 min
WORKFLOWRERUNSFAILS / TOTALDUR.
#1PR checks for TS39145 / 59414m
#2PR checks for Rust1495 / 6629m
#3Nightly030 / 303m
89% of re-runs from 5 of 29 workflows

Where is it coming from?

The pain is concentrated. A handful of workflows cause most of the re-runs and waiting - so a short fix list moves the whole org.

Descript merges 150,000+ tests without flaky retries
TRUST & SECURITY

Limited access to metadata, read only, and secure

Here is exactly what you grant when you connect

Trunk · CI Health Report
would like read-only access to:
READ-ONLY
Repository access · Read-only
  • MetadataRepo name, default branch, visibility✓ Read
  • ActionsWorkflow run history, pass/fail rates, timing✓ Read
  • Pull requestsCycle & merge time, first-review timestamps✓ Read
Never✗ Write access✗ Source code✗ Secrets
  • Read-only, enforced by GitHubA fine-grained GitHub App - not a legacy OAuth token. It can’t write to your repos.
  • Revoke in one clickGitHub → Settings → Applications → revoke. No email, no sales call, no “contact us.”
  • Delete your data anytimeAsk us and we remove everything on our end - no questions, no retention games.
  • Start with a single repoConnect just your busiest repo first. Expand scope later only if you want to.
HOW IT WORKS

Expert-reviewed, contextual, and actionable

The team at Trunk reads your numbers before you do. Here’s the whole flow.

01

Connect read-only

Authorize the GitHub App on the repos you choose. No test-code changes, no uploads, no agent to install.

02

We audit & benchmark

The team at Trunk reviews your real metrics by hand and benchmarks them against industry data and peer teams.

03

You get your report

A personalized breakdown of your reliability, wasted time, and worst offenders - plus an optional live walkthrough.

Stop guessing how healthy your CI is. Get the numbers.

Free, expert-reviewed, and read-only. We never read your source code - only your CI metadata.

Read-only metadata access, authorized by a GitHub admin. See exactly what you grant →