Costs and limits
Resources include Workers requests, Workers AI planning inference, Durable Object storage/execution, Containers compute, coding inference through the supported AI Gateway/provider arrangement, and GitHub API quotas.
No prices or free-use guarantees are stated here. Check Workers pricing, Durable Objects pricing, Containers pricing, Workers AI pricing, and AI Gateway.
Application limits
Section titled “Application limits”| Limit | Value |
|---|---|
| Maximum concurrent coding runs / configured container instances | 5 |
| OpenCode timeout | 15 minutes |
| Git command timeout | 5 minutes |
| Collected diff | 120,000 characters |
| Transcript diff display | 20,000 characters |
| stderr tail | 8,000 characters |
| Captured files | 50 |
| Per-file captured contents | 100,000 characters |
| Total captured contents | 500,000 characters |
Sources: apps/backend/src/runs.ts (MAX_CONCURRENT_RUNS = 5), apps/backend/src/runtime.ts, and apps/backend/src/transcript.ts. These are application settings, not Cloudflare plan quotas or guaranteed in-memory read bounds. Capture truncation can make publication incomplete.
No measured price or end-to-end cloud usage is available in the dated VERIFICATION.md; the earlier cost-ratio language is intentionally omitted because it is not evidenced here. Set account budgets and alerts where supported. Clearing history does not stop resource billing, and cancellation requests best-effort cleanup. Verify resource lifecycle and actual charges in your account.