[Nabhag8848]

Summary

Follow-up to the Blaxel review on #44. Compared Modal/Daytona patterns and Blaxel’s own SDKs, then only shipped fixes that reduce Go↔TS drift without inventing unsupported native APIs.

Shipped

• Ports examples: stop using Blaxel-reserved 443/8080; use 3000/3001
• Go floors (match TS + Modal’s memory-floor pattern): memory ≥ 256 MiB; max-age TTL ≥ 5m; idle TTL ≥ 1m
• TS delete: return Deleted + terminal after success (same as Go Blaxel / Daytona intent)
• Idle disable: TS now rejects like Go (fail-loud; silent omit was wrong)
• Docs: region wording (Go requires Config.Region; TS accepts Config.region or BL_REGION); scope includes workspace; configuration setup link for Blaxel

Explicitly skipped (would add debt)

• Go create timeout: Go Blaxel SDK NewInstance has no create-timeout option (TS @blaxel/core does). Fake-mapping via context.WithTimeout would not cancel/delete like Blaxel’s creation timeout.
• TS standby mapping: @blaxel/core SandboxInstance exposes status but not runtime state (RUNNING/STANDBY). Reading private fields is debt; Go already maps correctly.
• Full <ForProvider> expansion across shared guides: large permanent surface; dedicated blaxel.mdx + examples already cover Blaxel.

Test plan

• GOWORK=off go test ./... in sdks/go/providers/blaxel
• npm test / npm run build for @sandbox-kit/blaxel
• Build Go + TS create-blaxel-with-ports examples