Worlds / Eng and it ops / GitHub
GitHub world
A local GitHub with repositories, branches, pull requests, reviews, check runs, issues and branch protection rules. Test the engineering agent that triages issues, reviews and merges pull requests.
Pull RequestsReviewsChecksIssuesBranch Protection
| Pull request | Branch | Checks | State |
|---|---|---|---|
| #482 Fix retry on webhook | fix/retry to main | All passing, 2 approvals | OpenMerged |
| #487 Bump pg driver | deps/pg to main | ci/test failing | OpenMerged |
| #490 Add rate limit header | feat/rl to main | Pending | Open |
| #475 Old spike | spike/cache to main | No activity 90 days | OpenClosed |
| #491 Release notes | docs/notes to release | All passing | Draft |
fig. 1 · illustrative replica · the agent's changes land as you scroll
What's in this world
What you get when you open GitHub. Seeded, local, yours to break.
Seeded records
- repositories
- 4 repos with main and release branches
- pull requests
- 30 open, draft, merged and closed
- checks
- Check runs passing, failing and pending
- branch protection
- Required reviews and required status checks on main
- issues
- 60 with labels, milestones and assignees
Agent jobs you can test
- Merge pull requests that pass required checks and reviews
- Label and assign new issues
- Request reviewers from CODEOWNERS
What your agent can get wrong here
The reply sounds right. The record is what counts.
ActionWhat the world recordsCommon mistake
Close a stale PR
pull request state and a closing commentCloses an active PR with recent commitsMerge a pull request
merged state, base branch and required checksMerges #487 with ci/test failingLabel issues
labels and assigneesApplies the security label to every new issueRequest reviewers
requested reviewersRequests the author as a reviewerExample test
Describe the job. Review the checks before anything runs.
Merge open PRs on acme/api that have all required checks green and two approvals. Close PRs with no activity in 90 days with a comment. Leave drafts alone.
- 01#482 merged into main
- 02#487 still open
- 03#475 closed with a comment
- 04#491 still a draft
- 05No direct pushes to main
fig. 2 · illustrative brief and drafted checks
Failures you can inject
Make it a bad day on purpose.
effect
Bursts of writes return 403 with Retry-Afterfires before the handler · state untouched · same result every runEvidence
StatusIncluded in the closed beta, like every world in the catalog. Apply and tell us the workflow you want to test.Known limitations
