OneBuck.dev
CI Watch
CI observability for $1 per repository
Connect a GitHub Actions or CircleCI repository with a read-only token and see where your CI time goes: queueing versus running, which workflows and jobs dominate, how often runs fail, which jobs are flaky, and where cache misses hurt. OneBuck.dev reads run and job metadata only — it never runs your jobs.
One slot connects one GitHub Actions or CircleCI repository and includes 90 days of retained runs and jobs. Disconnect at any time to free the slot.
What CI Watch measures
- CI and queue time — total wall clock spent running versus waiting for a runner or executor.
- Failure rate — failed, timed-out and errored runs as a share of all runs in the window.
- Slowest workflows and jobs — ranked by total time so you can target the biggest consumers.
- Flaky jobs — jobs that both pass and fail across runs.
- Retries and reruns — how often runs and jobs are retried.
- Cache behaviour — cache misses versus total job restores where the provider reports it.
- Deterministic recommendations — plain-language hints such as queue-dominant builds or hot jobs.
Current limits
| Providers | GitHub Actions and CircleCI |
|---|---|
| Ingest model | Read-only polling every 15 minutes; no webhooks required |
| Initial backfill | Up to 90 days of history on first connect |
| History retained | 90 days of runs and jobs |
| Windows | 24h, 7d, 30d and 90d, filtered from ingested history |
| Token scope | Read-only access to repository and CI metadata |
| Jobs executed | None — observation only |
The overview reports the real ingested window next to the selected range, so you always know whether a quiet chart means a quiet repository or a short history.
Setup
- Create a read-only token: a fine-grained GitHub token with repository read access, or a CircleCI personal API token.
- Connect the repository from the dashboard with its owner and name.
- Allocate the $1 repository slot. The first sync backfills up to 90 days within a few minutes.
- Use the 24h, 7d, 30d and 90d range selector to compare windows.
Replacing a token re-validates it with the provider before it is stored, and any sync failure is shown against the repository instead of leaving a stale "Synced" time.
Common use cases
- Work out whether slow builds are queueing for runners or actually running.
- Find the one job that consumes most of your CI budget.
- Identify flaky jobs that cause noisy reruns.
- Track whether a cache tweak reduced misses over the last 30 days.
- Watch failure rate before and after a workflow change.
FAQ
What counts as one $1 repository?
One slot covers one repository on one provider for one month and includes 90 days of retained runs and jobs. Two repositories use two slots.
Do you run or modify my pipelines?
No. CI Watch only reads run, workflow and job metadata with a read-only token. It never triggers, edits or reruns your jobs.
How much history is available?
Up to 90 days. The first sync backfills as much history as the provider returns cheaply, and the overview always shows the actual ingested start date so ranges are never silently empty.
What happens if my token is revoked?
Token updates are validated before they are saved, and the next failed sync is recorded against the repository, so the dashboard shows the error instead of a stale "Synced" time.
Which providers are supported?
GitHub Actions and CircleCI on GitHub-backed projects today. More providers are on the roadmap.
Watch your CI for $1
Connect a repository from the dashboard, or head back to the shop to see every service.