[vitest-pool-workers] fix: reset() now clears ratelimit binding state between tests by matingathani · Pull Request #14409 · cloudflare/workers-sdk

matingathani · 2026-06-24T10:00:38Z

Problem

reset() from cloudflare:test only called workerdUnsafe.deleteAllDurableObjects(). Ratelimit bindings (RATE_LIMITERS) store their bucket counts in a Map on an in-memory Ratelimit instance. That instance is never restarted between tests, so bucket counts bleed across test boundaries — a test that exhausts a ratelimit causes subsequent tests in the same file to see it as already exhausted.

Solution

The miniflare ratelimit extension (ratelimit.worker.ts) now:

Adds a reset() method to the Ratelimit class that clears buckets and resets epoch.
Maintains a module-level Set<Ratelimit> and writes it to globalThis.__cfRatelimitInstances__ so the reset helper can reach it without importing an internal module directly.

reset.ts in vitest-pool-workers now iterates that set (if present) and calls reset() on each instance after deleteAllDurableObjects(). The global is only populated when the ratelimit extension is loaded, so the check is a no-op for workers with no ratelimit bindings.

Test

The existing reset fixture (fixtures/vitest-pool-workers-examples/reset) is extended with a RATE_LIMITER binding and two new tests:

"exhausts ratelimit then resets between tests" — exhausts the 100-request limit in one test
"sees reset ratelimit state after reset" — verifies the first call in the next test succeeds again

Tests
- Tests included/updated
Public documentation
- Documentation not necessary because: internal behaviour fix for existing reset() API, no new user-facing API surface

… between tests Fixes cloudflare#14392. The `reset()` helper from `cloudflare:test` only called `deleteAllDurableObjects()`, leaving ratelimit bucket counts intact across test boundaries. Subsequent tests in the same file could see stale exhaustion state from an earlier test. The fix uses a module-level registry in the miniflare ratelimit extension: each `Ratelimit` instance registers itself on `globalThis.__cfRatelimitInstances__` when created. `reset()` iterates that set and calls the new `Ratelimit.reset()` method, which clears `buckets` and resets `epoch`. The global is only populated when ratelimit bindings are actually configured, so the check is a no-op for workers without ratelimits. A test fixture (`fixtures/vitest-pool-workers-examples/reset`) is updated to verify that exhausting a ratelimit in one test does not bleed into the next.

Copilot

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

changeset-bot · 2026-06-24T10:00:47Z

🦋 Changeset detected

Latest commit: c78da86

The changes in this PR will be included in the next version bump.

This PR includes changesets to release 6 packages

Name	Type
miniflare	Patch
@cloudflare/vitest-pool-workers	Patch
@cloudflare/deploy-helpers	Patch
@cloudflare/pages-shared	Patch
@cloudflare/vite-plugin	Patch
wrangler	Patch

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR

workers-devprod · 2026-06-24T10:01:06Z

Codeowners approval required for this PR:

@cloudflare/wrangler

Show detailed file reviewers

.changeset/vitest-pool-workers-reset-ratelimit.md: [@cloudflare/wrangler]
fixtures/vitest-pool-workers-examples/reset/src/env.d.ts: [@cloudflare/wrangler]
fixtures/vitest-pool-workers-examples/reset/test/reset.test.ts: [@cloudflare/wrangler]
fixtures/vitest-pool-workers-examples/reset/wrangler.jsonc: [@cloudflare/wrangler]
packages/miniflare/src/workers/ratelimit/ratelimit.worker.ts: [@cloudflare/wrangler]
packages/vitest-pool-workers/src/worker/reset.ts: [@cloudflare/wrangler]

pkg-pr-new · 2026-06-24T10:03:53Z

@cloudflare/autoconfig

npm i https://pkg.pr.new/@cloudflare/autoconfig@14409

create-cloudflare

npm i https://pkg.pr.new/create-cloudflare@14409

@cloudflare/deploy-helpers

npm i https://pkg.pr.new/@cloudflare/deploy-helpers@14409

@cloudflare/kv-asset-handler

npm i https://pkg.pr.new/@cloudflare/kv-asset-handler@14409

miniflare

npm i https://pkg.pr.new/miniflare@14409

@cloudflare/pages-shared

npm i https://pkg.pr.new/@cloudflare/pages-shared@14409

@cloudflare/unenv-preset

npm i https://pkg.pr.new/@cloudflare/unenv-preset@14409

@cloudflare/vite-plugin

npm i https://pkg.pr.new/@cloudflare/vite-plugin@14409

@cloudflare/vitest-pool-workers

npm i https://pkg.pr.new/@cloudflare/vitest-pool-workers@14409

@cloudflare/workers-auth

npm i https://pkg.pr.new/@cloudflare/workers-auth@14409

@cloudflare/workers-editor-shared

npm i https://pkg.pr.new/@cloudflare/workers-editor-shared@14409

@cloudflare/workers-utils

npm i https://pkg.pr.new/@cloudflare/workers-utils@14409

wrangler

npm i https://pkg.pr.new/wrangler@14409

commit: c78da86

Co-authored-by: devin-ai-integration[bot] <158243242+devin-ai-integration[bot]@users.noreply.github.com>

…approach - Add miniflare: patch to changeset — without it the feature silently doesn't work because vitest-pool-workers pins to a miniflare version that lacks the instance registry and reset() method - Refactor the globalThis key from an instances Set to a single reset function reference (__cfRatelimitReset__) to minimise what's exposed on globalThis - The ratelimit extension module is marked `internal: true` in workerd, which prevents dynamic import from reset.ts; globalThis is the only cross-module communication path available within the same Worker isolate

petebacondarwin

Using a global to track instances feels very hacky.
Can we try to find a more elegant solution?

workers-devprod · 2026-06-29T10:09:27Z

Codeowners approval required for this PR:

@cloudflare/wrangler

Show detailed file reviewers

.changeset/vitest-pool-workers-reset-ratelimit.md: [@cloudflare/wrangler]
fixtures/vitest-pool-workers-examples/reset/src/env.d.ts: [@cloudflare/wrangler]
fixtures/vitest-pool-workers-examples/reset/test/reset.test.ts: [@cloudflare/wrangler]
fixtures/vitest-pool-workers-examples/reset/wrangler.jsonc: [@cloudflare/wrangler]
packages/miniflare/src/plugins/ratelimit/index.ts: [@cloudflare/wrangler]
packages/miniflare/src/workers/ratelimit/ratelimit-object.worker.ts: [@cloudflare/wrangler]
packages/miniflare/src/workers/ratelimit/ratelimit.worker.ts: [@cloudflare/wrangler]
packages/miniflare/test/plugins/local-explorer/r2.spec.ts: [@cloudflare/wrangler]
packages/miniflare/test/test-shared/miniflare.ts: [@cloudflare/wrangler]
packages/wrangler/e2e/dev.test.ts: [@cloudflare/wrangler]

petebacondarwin · 2026-06-30T09:35:57Z

@petebacondarwin The globalThis bridge is needed because the ratelimit extension module is marked internal: true in workerd — it can't be imported from reset.ts. The cleanest alternative I can think of is adding a resetRatelimitBindings() to workerd:unsafe (similar to deleteAllDurableObjects()), but that requires a workerd change. Do you have a specific pattern in mind?

If fixing this inside workerd is the best solution then we should do that rather than working around it - we do effectively own both parts of the system. I did briefly look for the rate limiting code in workerd but wasn't sure where it was found.

…bol.for for ratelimit reset Symbol.for("cloudflare:miniflare:ratelimit:reset") is collision-safe by construction and documents the same-isolate semantics: internal extension modules run inside the calling worker's V8 isolate, so globalThis is legitimately shared between ratelimit.worker.ts and reset.ts. Claude-Session: https://claude.ai/code/session_011QMWU4d8zZN3ERqCHrJz5i

…ing instead of globalThis Removes the globalThis[Symbol.for(...)] side-channel between the ratelimit extension module and reset(). Instead, the ratelimit plugin registers a reserved control binding (__MINIFLARE_RATELIMIT_CONTROL__) using the same wrapped-binding mechanism RATE_LIMITER itself already uses, and reset() calls resetAll() on it directly through env.

…ad of globalThis Removes the globalThis[Symbol.for(...)] bridge between the ratelimit extension module and reset(). The Ratelimit class already exposes a public reset() method; reset() now calls it directly on each configured ratelimit binding, using binding names read from the parsed wrangler config (pool/index.ts) via a new __VITEST_POOL_WORKERS_RATELIMIT_BINDING_NAMES module. This avoids walking `env` (which would touch RPC/entrypoint bindings like SELF) and avoids the miniflare wrapped-binding module entirely, so it needs no changes to the ratelimit plugin. Also fixes formatting on the existing changeset file (oxfmt).

matingathani · 2026-07-01T05:57:09Z

Pushed an update that removes the global entirely.

What changed

ratelimit.worker.ts no longer touches globalThis at all — the Ratelimit class just keeps its plain reset() method, nothing else. The miniflare ratelimit plugin is untouched.

The actual bridge moved into vitest-pool-workers, which already has everything it needs without reaching into Miniflare internals:

pool/index.ts reads the ratelimit binding names directly from the parsed wrangler config (same place RATE_LIMITER: RateLimit itself comes from) and injects them into the worker as a small module, __VITEST_POOL_WORKERS_RATELIMIT_BINDING_NAMES.

reset.ts imports that list and calls .reset() directly on each named binding via env:

import { env } from "cloudflare:workers";
import ratelimitBindingNames from "__VITEST_POOL_WORKERS_RATELIMIT_BINDING_NAMES";

for (const name of ratelimitBindingNames) {
  (env[name] as { reset?: () => void }).reset?.();
}

Since env.RATE_LIMITER already is the Ratelimit instance (the binding is just that object, wrapped), .reset() is a normal method call on a binding we already know by name — no globalThis, no shared internal module, no new binding, no touching Miniflare's extension/wrapped-binding machinery at all.

Why not the other things I tried

A dedicated "control" binding sharing the ratelimit extension's moduleName (my previous push) — workerd rejects two wrapped bindings pointing at the same internal module (wrapped binding module can't be resolved (internal modules only)), and it broke dev-server startup broadly, not just ratelimit-using workers. Reverted.
Walking Object.values(env) and duck-typing for .reset() — touches every binding including SELF/entrypoint-style ones, and property access on those triggers a real RPC handshake, throwing Expected default export ... to be a subclass of WorkerEntrypoint for RPC. The binding-name approach above avoids this by only ever touching bindings we already know are ratelimits.

Verification

Ran the full vitest-pool-workers-examples suite locally (56 test files / 160 tests, including reset, rpc, workflows, assets) — all green, no unhandled errors. pnpm check:type clean for both miniflare and vitest-pool-workers. CI is running now.

__VITEST_POOL_WORKERS_RATELIMIT_BINDING_NAMES was a virtual module from the original approach (vitest-pool-workers tracking ratelimit binding names for reset()). The DO-backed redesign removed that mechanism entirely; this tsdown external entry was the one leftover reference.

petebacondarwin

Thanks for this — the Durable-Object-backed redesign is the right call and it's cleanly done. Moving the bucket/epoch state into a MiniflareDurableObject so that deleteAllDurableObjects() (and therefore reset()) tears it down for free is exactly the pattern the other stateful plugins use, and the structure is near-identical to D1/KV (wrapped client + fetcher service inner-binding → per-binding object-entry worker → one DO-hosting service). Keeping option validation in the client isolate so errors still surface synchronously to the caller is a nice touch.

A few things I'd like to see addressed before merge:

1. In-memory DO is missing `preventEviction: true`

packages/miniflare/src/plugins/ratelimit/index.ts — the closest precedent is the Queues broker, which is also durableObjectStorage: { inMemory: kVoid } and sets preventEviction: true on the namespace (see packages/miniflare/src/plugins/queues/index.ts:97-105). Without it, the rate-limiter DO can be evicted while idle, which silently clears #buckets/#epoch mid-period. Please add preventEviction: true, and confirm it doesn't interfere with deleteAllDurableObjects() clearing state — it shouldn't, since that's a forced disposal, but it's worth verifying given that clearing is the whole point of this PR.

2. DO-hosting service is missing the standard object bindings

Both Queues and KV bind SharedBindings.MAYBE_SERVICE_LOOPBACK → SERVICE_LOOPBACK and spread getMiniflareObjectBindings(unsafeStickyBlobs) into the DO-hosting worker (queues/index.ts:106-111, kv/index.ts:167-177). The new RATELIMIT_ENTRY_SERVICE_PREFIX service has no bindings at all. RateLimiterObject only touches in-memory Maps today so it may work as-is, but the MiniflareDurableObject base machinery can reach for the loopback/blob services on some paths — please confirm they're genuinely unused here, or align with the Queues pattern to be safe.

3. Rate-limiter identity is the binding name, not `namespace_id` (I'll handle this in a follow-up)

getBindings/getServices key the object-entry service and the DO instance by the binding name. In production — and in the KV/D1 plugins, which key by id — a limiter is identified by namespace_id, so two bindings that share a namespace_id should share state, and two Workers that happen to use the same binding name should not collide. This is a pre-existing divergence that your redesign now makes straightforward to fix. I'm going to take this in a follow-up PR (key by namespace_id, and re-enable the multiworker ratelimits that are currently deleted in MultiworkerRuntimeController), so it does not block this PR — just flagging it here for context.

4. Unrelated changes bundled in

The Windows ECONNRESET retry helper (dispatchFetchWithRetry + the r2.spec.ts switch) and the EPIPE addition to ignoreEconnreset in e2e/dev.test.ts look like separate CI-flake fixes, unrelated to reset()/ratelimit. Could you split them into their own PR (or at least call them out in the description)? Keeps this one focused and easy to revert.

5. Changeset only bumps `miniflare`

reset() is the user-facing surface and it's re-exported from @cloudflare/vitest-pool-workers, but only miniflare gets a changeset entry. Consider adding a @cloudflare/vitest-pool-workers changeset too, so the fix shows up in the changelog where users of cloudflare:test will actually look.

Minor

The two new fixture tests are order-dependent (the second assumes the first exhausted the limit and that reset() ran in between). That's consistent with the existing Counter/KV/R2 cases in reset.test.ts, so I'm fine with it — just noting it.
The DO now takes limit/period per request, so the instance is config-agnostic. That's fine while it's keyed per-binding, but it interacts with the namespace_id change in (3); I'll account for it in the follow-up.

Overall this is close — happy to approve once 1, 2, 4, and 5 are sorted. (3 is on me.)

petebacondarwin · 2026-07-03T11:49:01Z

I just looked into workerd and it seems that preventEviction will indeed mean that deleteAllDurableObjects() will skip that DO. So this is probably not going to work...

matingathani · 2026-07-03T11:51:18Z

So should i go with 1,2,4 and 5 or wait for it ?

petebacondarwin · 2026-07-03T11:52:18Z

If the epoch is short enough, then perhaps the DO eviction period will be longer and we don't need preventEviction...

If that doesn't work out, we will still need to turn on eviction prevention, but then provide a direct API to that Rate Limiter DO that does this.ctx.storage.deleteAll() internally.

petebacondarwin · 2026-07-03T11:55:31Z

So should i go with 1,2,4 and 5 or wait for it ?

Let's do 2 & 5 now.
I am on the fence whether you need to pull the fixes (4) out into a separate PR. Your call.

matingathani · 2026-07-03T11:55:44Z

Ok , let me have a look at it and i will get back with a fix

…rdown handler" This reverts commit 220a54d.

… ECONNRESET" This reverts commit 322c480.

…hosting service Every other DO-hosting plugin (queues, kv, d1, r2) binds SharedBindings.MAYBE_SERVICE_LOOPBACK and spreads getMiniflareObjectBindings(unsafeStickyBlobs) into its object-entry service; the ratelimit plugin's RATELIMIT_ENTRY_SERVICE_PREFIX service had no bindings at all. Align it with the established pattern in case MiniflareDurableObject's base machinery reaches for the loopback/blob services on some path. Also add a changeset entry for @cloudflare/vitest-pool-workers, since reset() is the user-facing surface re-exported from that package. Addresses review items 2 and 5 from petebacondarwin on cloudflare#14409.

matingathani · 2026-07-06T07:22:20Z

Addressed 2, 4, 5 — leaving 1 (preventEviction/deleteAllDurableObjects interaction) and 3 (namespace_id keying, #14537) with you as discussed.

2: RATELIMIT_ENTRY_SERVICE_PREFIX's worker definition now binds SharedBindings.MAYBE_SERVICE_LOOPBACK → SERVICE_LOOPBACK and spreads getMiniflareObjectBindings(unsafeStickyBlobs), matching Queues/KV.
4: split the Windows ECONNRESET retry (dispatchFetchWithRetry) and hyperdrive e2e EPIPE-ignore fixes into their own PR — [miniflare][wrangler] Fix Windows CI flakes: local-explorer ECONNRESET retry, hyperdrive e2e EPIPE #14568. Reverted both commits from this branch.
5: .changeset/vitest-pool-workers-reset-ratelimit.md now also bumps @cloudflare/vitest-pool-workers.

Verified locally: packages/miniflare/test/plugins/ratelimit/index.spec.ts (2/2), the reset vitest-pool-workers-examples fixture (10/10, including the ratelimit reset case), and pnpm --filter miniflare check:type clean.

workers-devprod · 2026-07-06T07:22:38Z

Codeowners approval required for this PR:

✅ @cloudflare/wrangler

Show detailed file reviewers

RatelimitConfigSchema's period was optional with no default, so an omitted period reached buildJsonBindings() as undefined, producing an invalid `json: undefined` Worker_Binding instead of a valid JSON string — the client's RatelimitConfig.period is typed as a required number. Wrangler's own config validation already requires period for real wrangler.jsonc configs, but callers using the Miniflare API directly could hit this. Default it in the schema instead. Devin review finding on cloudflare#14409.

petebacondarwin · 2026-07-06T12:28:18Z

Eviction of DOs is done after 10 seconds of inactivity. But I don't think that any rate limiter would be expected to rate limit (in local dev) if there are fewer than one request every 10 secs... So I think this is a safe implementation without adding preventEviction: true.

If we get reports of this causing a problem we can also persist the DO state so that if the DO gets evicted it can reload it again afterwards.

workers-devprod

Codeowners reviews satisfied

petebacondarwin · 2026-07-06T12:29:16Z

I'll follow up with my PRs

matingathani · 2026-07-06T12:40:02Z

Sounds good , let me know if you any other issue persist , i will open up new pr to solve it

matingathani requested a review from workers-devprod as a code owner June 24, 2026 10:00

Copilot AI review requested due to automatic review settings June 24, 2026 10:00

Copilot AI reviewed Jun 24, 2026

workers-devprod added this to workers-sdk Jun 24, 2026

github-project-automation Bot moved this to Untriaged in workers-sdk Jun 24, 2026

workers-devprod requested review from a team and ascorbic and removed request for a team June 24, 2026 10:01

This comment was marked as resolved.

Sign in to view

matingathani and others added 2 commits June 25, 2026 11:19

Update .changeset/vitest-pool-workers-reset-ratelimit.md

633e556

Co-authored-by: devin-ai-integration[bot] <158243242+devin-ai-integration[bot]@users.noreply.github.com>

This comment was marked as resolved.

Sign in to view

petebacondarwin requested changes Jun 29, 2026

View reviewed changes

Comment thread fixtures/vitest-pool-workers-examples/reset/src/env.d.ts Outdated

github-project-automation Bot moved this from Untriaged to In Review in workers-sdk Jun 29, 2026

petebacondarwin marked this pull request as draft June 29, 2026 11:51

matingathani and others added 2 commits June 30, 2026 11:08

chore: regenerate env.d.ts with updated wrangler types hash

8500364

Merge branch 'main' into fix/vitest-pool-workers-reset-ratelimit

52edb7e

matingathani and others added 4 commits June 30, 2026 16:22

Merge branch 'main' into fix/vitest-pool-workers-reset-ratelimit

b449326

matingathani force-pushed the fix/vitest-pool-workers-reset-ratelimit branch from c4ca21e to e2f2246 Compare July 1, 2026 05:53

matingathani requested a review from petebacondarwin July 1, 2026 06:14

matingathani marked this pull request as ready for review July 3, 2026 10:00

matingathani requested a review from petebacondarwin July 3, 2026 10:05

petebacondarwin reviewed Jul 3, 2026

View reviewed changes

petebacondarwin mentioned this pull request Jul 3, 2026

[miniflare] Key local rate limit counters by namespace_id #14537

Closed

5 tasks

matingathani mentioned this pull request Jul 6, 2026

[miniflare][wrangler] Fix Windows CI flakes: local-explorer ECONNRESET retry, hyperdrive e2e EPIPE #14568

Merged

5 tasks

matingathani added 3 commits July 6, 2026 12:48

Revert "fix(wrangler): also ignore EPIPE in hyperdrive e2e socket tea…

48d7c06

…rdown handler" This reverts commit 220a54d.

Revert "fix(miniflare): retry local-explorer dispatchFetch on Windows…

cf60dfb

… ECONNRESET" This reverts commit 322c480.

This comment was marked as resolved.

Sign in to view

matingathani requested a review from petebacondarwin July 6, 2026 07:55

petebacondarwin removed the request for review from ascorbic July 6, 2026 12:16

petebacondarwin approved these changes Jul 6, 2026

View reviewed changes

workers-devprod approved these changes Jul 6, 2026

View reviewed changes

github-project-automation Bot moved this from In Review to Approved in workers-sdk Jul 6, 2026

petebacondarwin merged commit 16fbf81 into cloudflare:main Jul 6, 2026
62 checks passed

github-project-automation Bot moved this from Approved to Done in workers-sdk Jul 6, 2026

workers-devprod mentioned this pull request Jul 6, 2026

Version Packages #14536

Merged

petebacondarwin mentioned this pull request Jul 6, 2026

[miniflare] Key local rate limit counters by namespace_id #14572

Merged

5 tasks

Uh oh!

Conversation

matingathani commented Jun 24, 2026 • edited by devin-ai-integration Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Problem

Solution

Test

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Uh oh!

changeset-bot Bot commented Jun 24, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

🦋 Changeset detected

Uh oh!

workers-devprod commented Jun 24, 2026

Uh oh!

pkg-pr-new Bot commented Jun 24, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

This comment was marked as resolved.

Uh oh!

This comment was marked as resolved.

Uh oh!

petebacondarwin left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

workers-devprod commented Jun 29, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

petebacondarwin commented Jun 30, 2026

Uh oh!

matingathani commented Jul 1, 2026

Uh oh!

petebacondarwin left a comment

Choose a reason for hiding this comment

1. In-memory DO is missing preventEviction: true

2. DO-hosting service is missing the standard object bindings

3. Rate-limiter identity is the binding name, not namespace_id (I'll handle this in a follow-up)

4. Unrelated changes bundled in

5. Changeset only bumps miniflare

Minor

Uh oh!

petebacondarwin commented Jul 3, 2026

Uh oh!

matingathani commented Jul 3, 2026

Uh oh!

petebacondarwin commented Jul 3, 2026

Uh oh!

petebacondarwin commented Jul 3, 2026

Uh oh!

matingathani commented Jul 3, 2026

Uh oh!

matingathani commented Jul 6, 2026

Uh oh!

workers-devprod commented Jul 6, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

This comment was marked as resolved.

Uh oh!

petebacondarwin commented Jul 6, 2026

Uh oh!

workers-devprod left a comment

Choose a reason for hiding this comment

Uh oh!

petebacondarwin commented Jul 6, 2026

Uh oh!

Uh oh!

matingathani commented Jul 6, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

matingathani commented Jun 24, 2026 •

edited by devin-ai-integration Bot

Loading

changeset-bot Bot commented Jun 24, 2026 •

edited

Loading

pkg-pr-new Bot commented Jun 24, 2026 •

edited

Loading

workers-devprod commented Jun 29, 2026 •

edited

Loading

1. In-memory DO is missing `preventEviction: true`

3. Rate-limiter identity is the binding name, not `namespace_id` (I'll handle this in a follow-up)

5. Changeset only bumps `miniflare`

workers-devprod commented Jul 6, 2026 •

edited

Loading