packages/openapi-slicer/BEFORE_AFTER.md
Before / after
Generated by pnpm slice from /private/tmp/claude-502/-Users-malek-gara-Projects-gitcafe-agent-kit/166e1e79-676e-4884-95ed-493f660a49ff/scratchpad/raw/openapi.json.
operationId
| before | after | operation |
|---|---|---|
getApiReposByOwnerByRepoPulls | listPullRequests | GET /api/repos/{owner}/{repo}/pulls/ |
getApiReposByOwnerByRepoPullsByPr | getPullRequest | GET /api/repos/{owner}/{repo}/pulls/{pr} |
postApiReposByOwnerByRepoPulls | createPullRequest | POST /api/repos/{owner}/{repo}/pulls/ |
postApiReposByOwnerByRepoPullsByPrMerge | mergePullRequest | POST /api/repos/{owner}/{repo}/pulls/{pr}/merge |
postApiReposByOwnerByRepoPullsByPrClose | closePullRequest | POST /api/repos/{owner}/{repo}/pulls/{pr}/close |
getApiReposByOwnerByRepoIssues | listIssues | GET /api/repos/{owner}/{repo}/issues/ |
postApiReposByOwnerByRepoIssues | createIssue | POST /api/repos/{owner}/{repo}/issues/ |
deleteApiReposByOwnerByRepoIssuesByIssueLock | unlockIssue | DELETE /api/repos/{owner}/{repo}/issues/{issue}/lock |
getApiRepos | listRepositories | GET /api/repos |
getApiReposByOwnerByRepo | getRepository | GET /api/repos/{owner}/{repo}/ |
getApiReposByOwnerByRepoActionsRuns | listRepoActionRuns | GET /api/repos/{owner}/{repo}/actions/runs |
postApiOrgsBySlugActionsRunnersRegistration-token | createOrgActionRunnerRegistrationToken | POST /api/orgs/{slug}/actions/runners/registration-token |
Longest original id: getApiReposByOwnerByRepoWebhooksByWebhookIdDeliveriesByDeliveryIdAttempts (73 chars) → listWebhookDeliveryAttempts.
REST operations renamed: 547 of 547; webhook event ids (webhooks.issueCreated) are kept as-is.
What an agent has to load
| artifact | chars | ≈ tokens (chars/4) |
|---|---|---|
| full spec | 2,092,751 | 523,188 |
| largest slice (Pull Requests) | 475,683 | 118,921 |
| INDEX.md (all 619 operations) | 58,915 | 14,729 |
ops/listPullRequests.json (one operation) | 44,677 | 11,169 |
| median single operation | 23,914 | 5,979 |
| median slice | 127,205 | 31,801 |
Token numbers are a chars/4 estimate, not a tokenizer count.