refactor: src/exports holds only re-exports; implementation is placed by plane by wmadden-electric · Pull Request #136 · prisma/composer

wmadden-electric · 2026-07-20T14:23:58Z

What

src/exports/*.ts now holds nothing but re-exports, and no implementation
module imports from exports/. All implementation moved out — 21 files across
10 packages.

This reverses a characteristic ADR-0035
had recorded as accepted: that an exports file could be the implementation home
and internal modules could import "up" into it. That shipped in #130 and is
rejected here. The ADR is amended in this PR — a permanent decision record
documenting the opposite of what the code does is worse than no record.

Why it matters

An exports file that can hold implementation eventually does, and then internal
modules import it — at which point the published surface becomes something the
package's own internals depend on. That had already happened: pg-warm-resource.ts
and prisma-next-migrate.ts imported ./exports/pg-connection.ts, and
descriptors/prisma-next.ts imported ../exports/prisma-next.ts. Those are gone.

Where implementation lives, and why the layout is plane-shaped

src/exports/<name>.ts   # thin re-export only — the surface
src/<name>.ts           # shared-plane implementation
src/control/**          # control-plane implementation
src/execution/**        # execution-plane implementation

This is forced, not stylistic. The plane globs must not overlap (the cruiser has
no glob precedence), and core, cron and streams classify their roots with
src/*.ts → shared. A control- or execution-plane module at the root would be
classified shared, and a per-file override would overlap that glob and put the
file in two plane groups. Subdirectories are the only non-overlapping option.
target and storage keep one glob per root file instead, because their roots
are genuinely mixed-plane.

The directories are named after the planes the cruiser already enforces
(ADR-0017/0028) rather than importing Prisma Next's core/ name — one
vocabulary, not two.

This is stricter than the pattern it adapts: Prisma Next's own
adapter-postgres/src/exports/column-types.ts carries ~185 lines of
implementation. The stricter rule was taken deliberately.

Proof

Two workspace sweeps, both run at the tip:

56 exports files checked — zero contain a function/class/const/let/
var/enum/interface/type declaration.
112 non-test modules scanned — zero import from exports/.

Tests are the one exemption: a test or fixture importing the public surface is
consuming it as a consumer would, which is what test-import-patterns.mdc
prescribes — and core's probe-core-authoring.ts exists precisely to bundle the
public barrel and assert it stays lean.

Invariants held

Every published subpath key is unchanged, for every package. The multi-pass
packages (cron/storage/streams) keep exports:false, their hand-maintained
maps, and their multi-pass configs; their package.json#exports and dist ENTRY
filenames are byte-identical, which is what their runtime import.meta.url
sibling resolution depends on. (Internal content-hashed chunks may rehash — the
accepted invariant from #130. Amusingly, storage's reverted to their pre-#130
hashes, since moving implementation back to the root restored the module IDs.)

Full suite green: build 28/28 · typecheck 58/58 · lint · lint:deps
(no violations, 366 modules / 642 deps; architecture-coverage: 167 files
classified, 213 specifiers — every edge visible) · test 48/48 ·
lint:rules:symlinks consistent.

Two traps worth knowing (now in the rule)

export * does not carry a default. Three *-service.ts files have one —
the runnable node build.module points at. A bare export * would have
silently dropped it.
A program has nothing to re-export. The *-entrypoint.ts files are
side-effect programs with zero exports; their shims import for effect, which
is safe only while no package declares sideEffects: false.

Both are documented in .agents/rules/exports-entrypoints.mdc.

Also here

The Drive records for this slice and #130 (.drive/projects/exports-entrypoint/)
— .drive/projects/ is tracked in this repo and six other project records are on
main; this one was missing. Its learnings.md is corrected where this slice
falsified it.

🤖 Generated with Claude Code

Implementation moves out of src/exports/ to the package root, where the 0-foundation/** directory glob already classifies it as shared, and each exports file becomes a re-export of the same symbol set. The within-package secret test now imports the implementation module directly. The published subpath keys are unchanged (./assertions, ./casts, ./secret) and the generated exports map is byte-identical. Signed-off-by: willbot <w.a.madden+machine@gmail.com> Signed-off-by: Will Madden <madden@prisma.io>

Implementation moves out of src/exports/: the shared `testing.ts` to the package root (covered by src/*.ts -> shared) and the control-plane `deploy.ts` and `app-config.ts` to a new src/control/ directory. Each exports file becomes a re-export of the same symbol set; exports/index.ts was already thin and is unchanged. Control-plane implementation cannot sit at the root: core collapsed its root to src/*.ts -> shared, and a per-file override would overlap that glob, which the cruiser resolves by putting the file in both planes. A subdirectory keeps the globs non-overlapping, so architecture.config.json gains core/src/control/** -> control. deploy and app-config are mutually dependent and stay siblings, so their ./ imports of each other are unchanged. Within-package tests now import the implementation modules rather than the exports shims. The published subpath keys are unchanged (., ./config, ./deploy, ./testing) and the generated exports map is byte-identical. Signed-off-by: willbot <w.a.madden+machine@gmail.com> Signed-off-by: Will Madden <madden@prisma.io>

The `node()` build adapter moves to src/node.ts (shared) and the control-plane descriptor and assembler to src/control/build.ts, each exports file becoming a re-export of the same symbol set — including the default export the adapter ships. control/build.ts took the sibling type import with it, now reading the implementation at ../node.ts rather than the exports shim; its deploy-time esbuild build() call is untouched. architecture.config.json gains node/src/*.ts -> shared and node/src/control/** -> control; the two existing per-file exports globs are unchanged and nothing overlaps. The firewall test still walks from the authoring entry, so it now follows the shim into the implementation; its offending-import pattern is widened to catch the new src/control/ directory form as well as a /control entry. The published subpath keys are unchanged (., ./control) and the generated exports map is byte-identical. Signed-off-by: willbot <w.a.madden+machine@gmail.com> Signed-off-by: Will Madden <madden@prisma.io>

Mirrors the node package: the `nextjs()` build adapter moves to src/nextjs.ts (shared) and the control-plane descriptor and assembler to src/control/build.ts, each exports file becoming a re-export of the same symbol set including the default export. control/build.ts reads its sibling type from ../nextjs.ts rather than the exports shim; its deploy-time esbuild build() call is untouched. architecture.config.json gains nextjs/src/*.ts -> shared and nextjs/src/control/** -> control, non-overlapping with the existing per-file exports globs. The firewall test's offending-import pattern is widened to catch the src/control/ directory form. The published subpath keys are unchanged (., ./control) and the generated exports map is byte-identical. Signed-off-by: willbot <w.a.madden+machine@gmail.com> Signed-off-by: Will Madden <madden@prisma.io>

The deployment report renderer moves to src/render-deployment.ts and the exports file becomes a re-export. cli is classified by a single src/** -> control glob, which already covers both the package root and src/exports/, so architecture.config.json needs no change. The within-package test now imports the implementation module it exercises. The published subpath keys are unchanged (. and ./report) and the generated exports map is byte-identical. Signed-off-by: willbot <w.a.madden+machine@gmail.com> Signed-off-by: Will Madden <madden@prisma.io>

The Providers collection and the providers() bundle move to src/providers.ts; exports/index.ts keeps only re-exports — the implementation, the client, container and credential helpers, and the sibling compute and postgres entrypoints. An exports file re-exporting another exports file is fine, both being public surface; providers.ts imports only internal modules, never src/exports/. lowering is classified by a single src/** -> control glob, which already covers the package root, so architecture.config.json needs no change. The published subpath keys are unchanged (., ./compute, ./postgres, ./state) and the generated exports map is byte-identical. Signed-off-by: willbot <w.a.madden+machine@gmail.com> Signed-off-by: Will Madden <madden@prisma.io>

The control-plane extension entry moves to src/control/extension.ts — named for what it holds, the prismaCloud() ExtensionDescriptor factory — and the three shared surfaces (pg-connection, prisma-next, testing) move to the package root. Each exports file becomes a re-export; exports/index.ts was already thin and re-exports only non-moving root modules, so it is unchanged. This removes the inversion that motivated the rule: pg-warm-resource.ts and prisma-next-migrate.ts imported ./exports/pg-connection.ts and descriptors/prisma-next.ts imported ../exports/prisma-next.ts. All three now read the root implementation, so no implementation depends on the public surface. Tests may still import the surface, which is what they are for. target stays per-file in architecture.config.json — its root is mixed-plane, so src/*.ts would misclassify the control files there. It gains src/control/** -> control plus shared globs for the three new root modules; every existing glob is untouched and nothing overlaps. The invariants test's process-env inventory follows the implementation from exports/control.ts to control/extension.ts. The published subpath keys are unchanged (., ./connection, ./control, ./prisma-next, ./testing). Signed-off-by: willbot <w.a.madden+machine@gmail.com> Signed-off-by: Will Madden <madden@prisma.io>

The bare scheduler node and the scheduler boot program move to src/execution/, the plane their globs already named, and each exports file becomes a re-export. scheduler-service ships a default export, so its shim re-exports `default` explicitly rather than using `export *`, which would not carry it. scheduler-entrypoint exports nothing — it is a program — so its shim imports it for effect; the package declares no `sideEffects: false`, so that import is not dropped. src/execution/ is the same depth as src/exports/, so both moved files keep their ../scheduler.ts imports unchanged. architecture.config.json gains cron/src/execution/** -> execution; the existing globs are untouched. The exports map and every dist entry filename are byte-identical, which is what the runtime import.meta.url sibling resolution depends on. Signed-off-by: willbot <w.a.madden+machine@gmail.com> Signed-off-by: Will Madden <madden@prisma.io>

The storage service node moves to the package root as shared-plane implementation and the boot program to src/execution/, matching the planes their globs already named. Each exports file becomes a re-export; exports/index.ts and exports/testing.ts were already thin and are unchanged. storage-service ships a default export, so its shim re-exports both the named factory and `default`; storage-entrypoint exports nothing and is imported for effect. This removes the inversion: storage-module.ts imported ./exports/storage-service.ts and now reads the root implementation, so no implementation depends on the public surface. storage stays per-file in architecture.config.json — its root is mixed-plane, so src/*.ts would misclassify the runtime engine there. It gains src/execution/** -> execution and src/storage-service.ts -> shared. The exports map and every dist entry filename are byte-identical. The internal content-hashed chunks changed name, which is expected and allowed: moving the implementation back to the package root restored the module ids those hashes derive from. Runtime resolution targets the entry files, which did not move. Signed-off-by: willbot <w.a.madden+machine@gmail.com> Signed-off-by: Will Madden <madden@prisma.io>

The streams service node moves to the package root, where the collapsed src/*.ts -> shared glob already classifies it, and the boot program and the local stand-in move to src/execution/. testing.ts goes to src/execution/ rather than the root deliberately: it is execution-plane, and the root is collapsed to shared, which would misclassify it. Each exports file becomes a re-export. streams-service ships a default export, so its shim re-exports both the named factory and `default`; streams-entrypoint exports nothing and is imported for effect. This removes the inversion: streams-module.ts imported ./exports/streams-service.ts and now reads the root implementation. architecture.config.json gains streams/src/execution/** -> execution; the existing globs are untouched and nothing overlaps. The exports map and every dist entry filename are byte-identical; only the internal content-hashed chunks changed name, which is allowed. The entrypoint integration test, which boots the entry as a child process, still passes. Signed-off-by: willbot <w.a.madden+machine@gmail.com> Signed-off-by: Will Madden <madden@prisma.io>

…placed by plane ADR-0035 recorded the opposite of what the workspace now does. It documented as an accepted characteristic that implementation may live in src/exports/ and that internal code may import "up" into an exports/ sibling, naming the stricter form an optional future refactor. The operator rejected that, and the workspace has been converted, so the ADR is corrected rather than left contradicting the code. It now records: an exports file holds nothing but re-exports; no implementation module may import from exports/, with tests exempt because a test importing the public surface is consuming it as a consumer would; and implementation is placed by plane — shared at the package root, control in src/control/**, execution in src/execution/**. That placement is forced rather than stylistic: packages whose root is collapsed to src/*.ts -> shared would misclassify a control or execution module placed there, and a per-file override would overlap the collapsed glob, which with no glob precedence puts the file in two plane groups. The ADR also now says plainly that this is stricter than the pattern it adapts, whose own adapter-postgres exports file carries implementation, and why we chose the stricter rule. The generated-map decision, the two hand-maintained exceptions, the entry-filename invariant, and the non-overlapping-globs constraint are unchanged. The rule gains the thin-surface requirement, the tests exemption, the plane placement, and the two shim traps this slice hit: `export *` does not carry a `default` export, and an entrypoint program with no exports is imported for effect. Signed-off-by: willbot <w.a.madden+machine@gmail.com> Signed-off-by: Will Madden <madden@prisma.io>

Adds the project record for the slice that moved every package's public entrypoints into src/exports/ (merged as PR #130) and for the follow-up that made those files thin: the investigation, the slice specs, the dispatch briefs, and the learnings. Matches how the six existing project records are stored; reviews/ stays untracked per .gitignore. Two learnings entries are corrected because later work falsified them. The "internal code imports up into src/exports/" entry recorded an accepted characteristic that the follow-up slice reversed — it is kept, marked as reversed, with the lesson that an exports file which can hold implementation eventually does, and that being green under the current rules is not evidence the shape is right. The "D6 collapse dropped" entry now records that D9 fixed the glob normalizer and took the collapse, with the lesson that a separate concern blocking a cleanup is worth pricing before dropping the cleanup. A new entry records this slice's traps: `export *` silently drops a `default` export, entrypoint programs have nothing to re-export so their shim imports for effect, implementation placement is forced by the plane globs, and a test that identifies a file by path breaks silently when the file moves. Signed-off-by: willbot <w.a.madden+machine@gmail.com> Signed-off-by: Will Madden <madden@prisma.io>

pkg-pr-new · 2026-07-20T14:25:22Z

Open in StackBlitz

npm i https://pkg.pr.new/@prisma/composer@136

npm i https://pkg.pr.new/@prisma/composer-prisma-cloud@136

commit: f04ddfc

Brings the branch up to date with main (postgres Branch naming in create, the cast-reading rules doc, the extension-registry model refresh, and the SPI-inversion project close-out). No conflicts: main touched no file this branch changed. Signed-off-by: willbot <w.a.madden+machine@gmail.com> Signed-off-by: Will Madden <madden@prisma.io>

Brings in main's bucket() resource (#132). The additive bucket files — new lowering providers (buckets/Bucket.ts, buckets/BucketKey.ts, exports/buckets.ts), the target descriptor and node module, the example app, and the CI job — merge cleanly. Two files conflicted because bucket added exports to what are, on this branch, thin re-export shims. Resolved on the slice's principle — main's content, our placement: - lowering/src/exports/index.ts: main folded the Bucket/BucketKey providers into providers() and re-exported ./buckets.ts. The providers() additions go to the implementation in src/providers.ts; the shim keeps only the new re-export. - target/src/exports/control.ts: main registered the bucket descriptor under the 's3' node kind inside prismaCloud(). That goes to the implementation in src/control/extension.ts; the shim stays `export * from '../control/extension.ts'`. Signed-off-by: willbot <w.a.madden+machine@gmail.com> Signed-off-by: Will Madden <madden@prisma.io>

coderabbitai · 2026-07-21T09:07:48Z

Warning

Review limit reached

@wmadden-electric, you've reached your PR review limit, so we couldn't start this review.

Next review available in: 16 minutes

Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available.
You're only billed for reviews past your plan's rate limits ($0.25/file).

How can I continue?

After more reviews become available, a review can be triggered using the @coderabbitai review command as a PR comment. Alternatively, push new commits to this PR.

To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews.

How do review limits work?

CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability.

For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window.

Please refer docs for additional details.

Review details

⚙️ Run configuration

Configuration used: Organization UI

Review profile: ASSERTIVE

Plan: Pro

Run ID: 5f3e67db-5070-44ca-847a-bbb00272fbe9

📥 Commits

Reviewing files that changed from the base of the PR and between 842c125 and f04ddfc.

📒 Files selected for processing (73)

.agents/rules/exports-entrypoints.mdc
.drive/projects/exports-entrypoint/investigation.md
.drive/projects/exports-entrypoint/learnings.md
.drive/projects/exports-entrypoint/slices/exports-entrypoint/d3-brief.md
.drive/projects/exports-entrypoint/slices/exports-entrypoint/d4a-brief.md
.drive/projects/exports-entrypoint/slices/exports-entrypoint/d4b-brief.md
.drive/projects/exports-entrypoint/slices/exports-entrypoint/d5-brief.md
.drive/projects/exports-entrypoint/slices/exports-entrypoint/d6-brief.md
.drive/projects/exports-entrypoint/slices/exports-entrypoint/d7-brief.md
.drive/projects/exports-entrypoint/slices/exports-entrypoint/d8-brief.md
.drive/projects/exports-entrypoint/slices/exports-entrypoint/spec.md
.drive/projects/exports-entrypoint/slices/thin-exports/spec.md
architecture.config.json
docs/design/90-decisions/ADR-0035-public-entrypoints-live-in-src-exports.md
packages/0-framework/0-foundation/foundation/src/assertions.ts
packages/0-framework/0-foundation/foundation/src/casts.ts
packages/0-framework/0-foundation/foundation/src/exports/assertions.ts
packages/0-framework/0-foundation/foundation/src/exports/casts.ts
packages/0-framework/0-foundation/foundation/src/exports/secret.ts
packages/0-framework/0-foundation/foundation/src/secret.test.ts
packages/0-framework/0-foundation/foundation/src/secret.ts
packages/0-framework/1-core/core/src/__tests__/lowering.test.ts
packages/0-framework/1-core/core/src/__tests__/testing.test-d.ts
packages/0-framework/1-core/core/src/__tests__/testing.test.ts
packages/0-framework/1-core/core/src/control/app-config.ts
packages/0-framework/1-core/core/src/control/deploy.ts
packages/0-framework/1-core/core/src/exports/app-config.ts
packages/0-framework/1-core/core/src/exports/deploy.ts
packages/0-framework/1-core/core/src/exports/testing.ts
packages/0-framework/1-core/core/src/testing.ts
packages/0-framework/2-authoring/nextjs/src/__tests__/no-control-import.test.ts
packages/0-framework/2-authoring/nextjs/src/control/build.ts
packages/0-framework/2-authoring/nextjs/src/exports/control.ts
packages/0-framework/2-authoring/nextjs/src/exports/index.ts
packages/0-framework/2-authoring/nextjs/src/nextjs.ts
packages/0-framework/2-authoring/node/src/__tests__/no-control-import.test.ts
packages/0-framework/2-authoring/node/src/control/build.ts
packages/0-framework/2-authoring/node/src/exports/control.ts
packages/0-framework/2-authoring/node/src/exports/index.ts
packages/0-framework/2-authoring/node/src/node.ts
packages/0-framework/3-tooling/cli/src/__tests__/render-deployment.test.ts
packages/0-framework/3-tooling/cli/src/exports/render-deployment.ts
packages/0-framework/3-tooling/cli/src/render-deployment.ts
packages/1-prisma-cloud/0-lowering/lowering/src/exports/index.ts
packages/1-prisma-cloud/0-lowering/lowering/src/providers.ts
packages/1-prisma-cloud/1-extensions/target/src/__tests__/invariants.test.ts
packages/1-prisma-cloud/1-extensions/target/src/control/extension.ts
packages/1-prisma-cloud/1-extensions/target/src/descriptors/prisma-next.ts
packages/1-prisma-cloud/1-extensions/target/src/exports/control.ts
packages/1-prisma-cloud/1-extensions/target/src/exports/pg-connection.ts
packages/1-prisma-cloud/1-extensions/target/src/exports/prisma-next.ts
packages/1-prisma-cloud/1-extensions/target/src/exports/testing.ts
packages/1-prisma-cloud/1-extensions/target/src/pg-connection.ts
packages/1-prisma-cloud/1-extensions/target/src/pg-warm-resource.ts
packages/1-prisma-cloud/1-extensions/target/src/prisma-next-migrate.ts
packages/1-prisma-cloud/1-extensions/target/src/prisma-next.ts
packages/1-prisma-cloud/1-extensions/target/src/testing.ts
packages/1-prisma-cloud/2-shared-modules/cron/src/execution/scheduler-entrypoint.ts
packages/1-prisma-cloud/2-shared-modules/cron/src/execution/scheduler-service.ts
packages/1-prisma-cloud/2-shared-modules/cron/src/exports/scheduler-entrypoint.ts
packages/1-prisma-cloud/2-shared-modules/cron/src/exports/scheduler-service.ts
packages/1-prisma-cloud/2-shared-modules/storage/src/execution/storage-entrypoint.ts
packages/1-prisma-cloud/2-shared-modules/storage/src/exports/storage-entrypoint.ts
packages/1-prisma-cloud/2-shared-modules/storage/src/exports/storage-service.ts
packages/1-prisma-cloud/2-shared-modules/storage/src/storage-module.ts
packages/1-prisma-cloud/2-shared-modules/storage/src/storage-service.ts
packages/1-prisma-cloud/2-shared-modules/streams/src/execution/streams-entrypoint.ts
packages/1-prisma-cloud/2-shared-modules/streams/src/execution/testing.ts
packages/1-prisma-cloud/2-shared-modules/streams/src/exports/streams-entrypoint.ts
packages/1-prisma-cloud/2-shared-modules/streams/src/exports/streams-service.ts
packages/1-prisma-cloud/2-shared-modules/streams/src/exports/testing.ts
packages/1-prisma-cloud/2-shared-modules/streams/src/streams-module.ts
packages/1-prisma-cloud/2-shared-modules/streams/src/streams-service.ts

✨ Finishing Touches

🧪 Generate unit tests (beta)

Create PR with unit tests
Commit unit tests in branch claude/exports-thin-surfaces

✨ Simplify code

Create PR with simplified code
Commit simplified code in branch claude/exports-thin-surfaces

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

_{Comment @coderabbitai help to get the list of available commands.}

Signed-off-by: willbot <w.a.madden+machine@gmail.com> Signed-off-by: Will Madden <madden@prisma.io>

wmadden-electric added 12 commits July 20, 2026 15:45

wmadden approved these changes Jul 21, 2026

View reviewed changes

wmadden-electric enabled auto-merge (squash) July 21, 2026 09:09

Merge origin/main into claude/exports-thin-surfaces

f04ddfc

Signed-off-by: willbot <w.a.madden+machine@gmail.com> Signed-off-by: Will Madden <madden@prisma.io>

wmadden-electric merged commit 621149f into main Jul 21, 2026
12 of 13 checks passed

wmadden-electric deleted the claude/exports-thin-surfaces branch July 21, 2026 09:12

wmadden-electric mentioned this pull request Jul 22, 2026

fix: invalid ids/names fail at authoring instead of mid-deploy; conformance suite repaired #149

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

refactor: src/exports holds only re-exports; implementation is placed by plane#136

refactor: src/exports holds only re-exports; implementation is placed by plane#136
wmadden-electric merged 15 commits into
mainfrom
claude/exports-thin-surfaces

wmadden-electric commented Jul 20, 2026

Uh oh!

pkg-pr-new Bot commented Jul 20, 2026 •

edited

Loading

Uh oh!

coderabbitai Bot commented Jul 21, 2026 •

edited

Loading

Review limit reached

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

wmadden-electric commented Jul 20, 2026

What

Why it matters

Where implementation lives, and why the layout is plane-shaped

Proof

Invariants held

Two traps worth knowing (now in the rule)

Also here

Uh oh!

pkg-pr-new Bot commented Jul 20, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

coderabbitai Bot commented Jul 21, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Review limit reached

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

pkg-pr-new Bot commented Jul 20, 2026 •

edited

Loading

coderabbitai Bot commented Jul 21, 2026 •

edited

Loading