[Brooooooklyn]

What

Adds worker.limits.cpu_ms to void.json so a project can set its own Workers CPU cap per request.

{
  "worker": {
    "limits": { "cpu_ms": 30000 }
  }
}

Context: platform plans cap Workers-for-Platforms cpuMs (pro = 50 ms today). WASM workloads such as image generation need more. This gives the project a knob, and the platform clamps it to a per-plan ceiling (platform PR follows).

Changes

• config.ts + schema.json: worker.limits validated as a strict object (unknown keys rejected), cpu_ms an integer in 1..300000 (Cloudflare's hard maximum)
• Deploy manifest: optional additive limits: { cpuMs } field; manifest version stays 3, so old CLI / old API pairs keep working
• Emitted by buildVoidManifest and buildFrameworkManifest; static manifests never carry limits
• The value is also copied into the Cloudflare vite-plugin worker config, so the generated dist/ssr/wrangler.json enforces it on void deploy --backend cloudflare (found by adversarial review: without this the knob silently no-ops on the self-hosted path)
• Docs: reference/config.md documents both paths; on the managed platform, a deploy above the plan ceiling rejects with 400, and only rollback clamps

Tests

• config.test.ts: validation cases plus a schema-drift pin asserting schema.json bounds match runtime validation
• package.test.ts: manifest emission for all three builders
• plugin-dev-asset-routing.test.ts: the merged Cloudflare worker config carries limits.cpu_ms when configured, and no limits key otherwise
• Full packages/void suite: 199 files passed, 3700 tests passed, 14 skipped (pre-existing gated live e2e)

🤖 Generated with Claude Code

https://claude.ai/code/session_01DF8p6MCmgiE7e3LM1J8EQV

---

Note

Medium Risk
Changes how deploy manifests and generated wrangler config express per-request CPU limits, which can affect runtime behavior and plan/account enforcement on production deploys, though values are strictly bounded and the manifest field is additive.

Overview
Projects can now set Workers CPU time per request via void.json under worker.limits.cpu_ms (integer 1–300000 ms), with strict validation, JSON schema support, and docs for managed vs self-hosted behavior.

The value flows into managed deploy manifests as optional limits.cpuMs (void and framework packaging paths; static manifests omit it) and into the generated Cloudflare worker config / dist/ssr/wrangler.json for void deploy --backend cloudflare and direct wrangler deploys. Deploy CLI passes the setting from config into both packaging flows.

Tests cover config validation, schema drift, manifest emission, and plugin worker config merging.

^{Reviewed by Cursor Bugbot for commit 59902dd. Bugbot is set up for automated code reviews on this repo. Configure here.}

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: c752d8694c

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".