Repository navigation
Implement repo-backed source foundation - #181
Conversation
|
Note Reviews pausedIt looks like this branch is under active development. To avoid overwhelming you with review comments due to an influx of new commits, CodeRabbit has automatically paused this review. You can configure this behavior by changing the Use the following commands to manage reviews:
Use the checkboxes below for quick actions:
📝 WalkthroughWalkthroughAdds repository-backed artifact/session infrastructure and MCP repo capabilities; persists and syncs entity sources; enables repo-backed execution for apps/skills/jobs; integrates jobs into vector search and unified search; introduces DB migrations, Durable Object repo sessions, maintenance reindex endpoint, and related types/tests/tooling. Changes
Sequence Diagram(s)sequenceDiagram
rect rgba(200,200,255,0.5)
actor User
participant JobSvc as Job Service
participant RepoRPC as RepoSession RPC
participant Shell as Cloudflare Shell (Workspace)
participant AppDB as D1 Database
participant AI as AI / Executor
end
User->>JobSvc: trigger job run (repo-backed)
JobSvc->>RepoRPC: openSession({ sourceId, userId, ... })
RepoRPC->>Shell: fork/clone repo, init workspace
RepoRPC->>AppDB: upsert repo_sessions
RepoRPC-->>JobSvc: sessionId
JobSvc->>RepoRPC: read entrypoint / runChecks
RepoRPC->>Shell: read files, run checks, compute tree/hash
RepoRPC->>AppDB: persist check results
RepoRPC-->>JobSvc: manifest + file content
JobSvc->>AI: run codemode with entrypoint + repoContext
AI-->>JobSvc: execution result
JobSvc-->>User: ExecuteResult
sequenceDiagram
rect rgba(200,255,200,0.5)
actor User
participant UISave as ui_save_app Handler
participant EntitySvc as Entity Source Service
participant Artifacts as ARTIFACTS Binding
participant RepoRPC as RepoSession RPC
participant AppDB as D1 Database
end
User->>UISave: save app with source intent
UISave->>EntitySvc: ensureEntitySource({ userId, entityKind:'app', entityId })
EntitySvc->>AppDB: query entity_sources
alt exists
EntitySvc-->>UISave: return existing source
else create
EntitySvc->>Artifacts: create/fork repo
Artifacts-->>EntitySvc: repo handle
EntitySvc->>AppDB: insert entity_sources
EntitySvc-->>UISave: created source
end
UISave->>AppDB: insert/update ui_artifact (with sourceId)
UISave->>RepoRPC: openSession({ sourceId, userId, ... })
RepoRPC->>Shell: apply edits (client/server/kody.json)
RepoRPC->>RepoRPC: publishSession -> push & update entity_sources.published_commit
RepoRPC-->>UISave: published commit
UISave->>AppDB: update ui_artifact.published_commit
Estimated code review effort🎯 5 (Critical) | ⏱️ ~120 minutes Possibly related PRs
Poem
✨ Finishing Touches🧪 Generate unit tests (beta)
|
![cursor[bot]](https://avatars.githubusercontent.com/in/1210556?s=60&v=4){kind=small}
|
🔎 Preview deployed: https://kody-pr-181.kentcdodds.workers.dev Worker: Mocks:
|
Co-authored-by: Kent C. Dodds <me+github@kentcdodds.com>
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 13
Note
Due to the large number of review comments, Critical, Major severity comments were prioritized as inline comments.
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (5)
packages/worker/src/mcp/capabilities/meta/meta-save-skill.ts (1)
199-241:⚠️ Potential issue | 🟡 MinorAdd
deleteEntitySourcecall to rollback path for new skill insert failures.When a new skill insert succeeds but
upsertSkillVectorfails, the recovery path deletes the skill viadeleteMcpSkillbut leaves theentity_sourcesrow created byensureEntitySource. On retry, a new UUID is generated for the skill (line 127:crypto.randomUUID()), soensureEntitySourcewon't reuse the orphaned row—it will create another one, leaking both the oldentity_sourcesentry and its synced artifacts in the repo. ImportdeleteEntitySourceand call it in theelseblock (line 235) beforedeleteMcpSkill:} else { + await deleteEntitySource(ctx.env.APP_DB, { + id: source.id, + userId: user.userId, + }) await deleteMcpSkill( ctx.env.APP_DB, user.userId, prep.rowPayload.name, ) }🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@packages/worker/src/mcp/capabilities/meta/meta-save-skill.ts` around lines 199 - 241, The rollback path after a failed upsertSkillVector leaves an orphaned entity_sources row created earlier by ensureEntitySource because retries generate a new skill id via crypto.randomUUID(); to fix, import deleteEntitySource and invoke deleteEntitySource(ctx.env.APP_DB, user.userId, prep.rowPayload.source_id or the source id used) in the else branch just before calling deleteMcpSkill in meta-save-skill.ts so the entity_sources row is removed alongside the MCP skill; ensure the call uses the same DB env and userId as deleteMcpSkill and matches the source identifier that ensureEntitySource created.packages/worker/src/mcp/capabilities/apps/ui-save-app.ts (1)
188-274:⚠️ Potential issue | 🟠 MajorRunner/vector failures leave the published source snapshot orphaned.
The update path publishes the source snapshot (
syncArtifactSourceSnapshot) beforeconfigureSavedAppRunnerruns insaveAndIndexApp. If runner configuration fails, the artifact row is rolled back toexistingAppvalues, but the repo commit recorded bysyncArtifactSourceSnapshotis not rolled back — the source now advertises a commit whose corresponding DB state was reverted. The insert path has the same issue: on failure we delete the artifact row and runner but leave the newly-synced source/commit behind.Given the PR's goal of making saved writes source-aware, consider either (a) reordering so runner+vector configuration happen before source sync/publish, or (b) tracking the synced commit and explicitly compensating (discard/rewind) on the error paths.
Also applies to: 340-358, 398-412
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@packages/worker/src/mcp/capabilities/apps/ui-save-app.ts` around lines 188 - 274, The source snapshot commit produced by syncArtifactSourceSnapshot can remain published when subsequent steps (configureSavedAppRunner, upsertUiArtifactVector) fail, leaving an orphaned repo commit; fix by either moving syncArtifactSourceSnapshot to run after configureSavedAppRunner/upsertUiArtifactVector or by capturing the commit id returned by syncArtifactSourceSnapshot and performing a compensating cleanup on error paths (the catch blocks that currently call deleteSavedAppRunner and deleteUiArtifact/deleteUiArtifactVector); specifically update the flows around configureSavedAppRunner, deleteSavedAppRunner, upsertUiArtifactVector, and deleteUiArtifactVector to either reorder operations or call a rollback API to discard/rewind the synced commit when runner/vector setup fails.packages/worker/src/jobs/service.ts (1)
96-122:⚠️ Potential issue | 🟡 MinorMisleading error message when both
codeandsourceIdare absent.Line 115 still reads
'Jobs require code.', but the type surface now accepts repo-backed jobs viasourceId. The check itself is correct; the user-facing message should reflect the new option so callers don't thinkcodeis the only valid update path.🐛 Proposed fix
- if (!nextCode && !nextSourceId) { - throw new Error('Jobs require code.') - } + if (!nextCode && !nextSourceId) { + throw new Error('Jobs require either code or a sourceId.') + }🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@packages/worker/src/jobs/service.ts` around lines 96 - 122, The error thrown in resolveUpdatedShape is misleading: when both nextCode and nextSourceId are absent it currently throws "Jobs require code." Update the thrown Error message in resolveUpdatedShape to mention that either code or sourceId is required (e.g., "Jobs require either code or sourceId.") so callers understand repo-backed jobs via sourceId are valid; locate the check that uses nextCode and nextSourceId and replace the message accordingly.packages/worker/src/mcp/tools/search.ts (1)
552-558:⚠️ Potential issue | 🟡 Minor
entityschema.describe(...)text is missingjob.The tool's prose description (Lines 127-138) was updated to announce
job, but the Zod schema's.describe(...)on Line 557 still lists onlycapability, skill, app, secret, value, or connector. Clients that introspect the input schema will see an inconsistent contract.🐛 Proposed fix
- 'Optional exact entity reference in the format "{id}:{type}" where type is capability, skill, app, secret, value, or connector.', + 'Optional exact entity reference in the format "{id}:{type}" where type is capability, skill, app, job, secret, value, or connector.',🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@packages/worker/src/mcp/tools/search.ts` around lines 552 - 558, The Zod schema for the entity field (the z.string().min(1).optional() chain for the entity property) has an outdated .describe(...) text that omits "job"; update the .describe(...) string on the entity schema to list "job" among the allowed types (e.g., "capability, skill, app, secret, value, connector, or job") so the schema prose matches the updated tool description and consumers that introspect the schema see a consistent contract.packages/worker/src/mcp/tools/search-format.ts (1)
278-282:⚠️ Potential issue | 🟡 MinorError message missing
jobin the accepted-types list.Line 295 was updated to include
job, but the format-error on Line 280 (thrown when:is missing/misplaced) still lists onlycapability, skill, app, secret, value, or connector. For consistency with the type union and the other error message, addjobhere too.🐛 Proposed fix
- 'Entity must use the format "{id}:{type}" where type is capability, skill, app, secret, value, or connector.', + 'Entity must use the format "{id}:{type}" where type is capability, skill, app, job, secret, value, or connector.',🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@packages/worker/src/mcp/tools/search-format.ts` around lines 278 - 282, The thrown Error inside the conditional "if (separator <= 0 || separator === trimmed.length - 1)" in search-format.ts lists accepted types but omits "job"; update that error string to match the updated union and other error messages by adding "job" to the list so it reads something like '{id}:{type}' where type is capability, skill, app, secret, value, connector, or job.
🟡 Minor comments (19)
packages/worker/wrangler.jsonc-8-13 (1)
8-13:⚠️ Potential issue | 🟡 MinorSchema is valid, but consider moving
artifactsinto eachenvblock for consistency.The
artifactsbinding schema is correct per Cloudflare documentation (binding, namespace fields, "default" namespace value all valid). However, all other bindings in this config (ai, worker_loaders, assets, durable_objects, d1_databases, kv_namespaces, vectorize) are defined withinenv.production,env.preview, andenv.testblocks. Placingartifactsat the top-level while everything else is environment-specific breaks the architectural pattern and may cause confusion during maintenance. Move the artifacts binding into each environment block for parity.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@packages/worker/wrangler.jsonc` around lines 8 - 13, Move the top-level "artifacts" binding into each environment block so it matches the pattern used for other bindings; remove the top-level "artifacts" array and add an "artifacts" entry with the same object ({ "binding": "ARTIFACTS", "namespace": "default" }) inside env.production, env.preview, and env.test (matching the existing structure used for ai, worker_loaders, assets, durable_objects, d1_databases, kv_namespaces, vectorize) so each env contains its own artifacts binding.packages/worker/src/agent-turn/tools.ts-65-69 (1)
65-69:⚠️ Potential issue | 🟡 MinorAdd a test for
loadJobsfailure to match coverage for other loaders.The
loadOptionalSearchRowsfunction properly isolatesloadJobserrors usingPromise.allSettled()and the warning pattern (lines 294–302 ofsearch.ts), matching the error handling for skills, artifacts, and values. However, the test file lacks a symmetric test forloadJobsfailure—tests exist for skills (line 43), apps (line 76), and values (line 96), but not for jobs. Add a test case to verify that aloadJobsrejection generates the"Saved jobs are temporarily unavailable"warning and doesn't poison the search result.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@packages/worker/src/agent-turn/tools.ts` around lines 65 - 69, Add a unit test that simulates `listJobs` (used by `loadJobs` inside `loadOptionalSearchRows`) rejecting and asserts that the rejection produces the warning message "Saved jobs are temporarily unavailable" and does not break the overall search result; mock `listJobs` to return a rejected promise, call `loadOptionalSearchRows` (or the search helper used by existing tests for skills/apps/values), then assert the warning was recorded (or the warnings array contains that string) and the search payload is still returned/usable. Ensure the test mirrors the pattern and assertions used in the existing skills/apps/values test cases so coverage and behavior are symmetric.packages/worker/src/job-maintenance.ts-20-22 (1)
20-22:⚠️ Potential issue | 🟡 MinorMinor: consider constant-time token comparison.
Bearer token is compared with
!==, which is not constant-time. For a maintenance endpoint behind a shared secret this is typically fine on Workers, but a timing-safe comparison (e.g., hashing both sides and comparing, or subtle/crypto-based) is a cheap hardening win if this endpoint is reachable from the public internet.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@packages/worker/src/job-maintenance.ts` around lines 20 - 22, The bearer token comparison using `token !== secret` is susceptible to timing attacks; replace it with a constant-time comparison: compute a fixed-length digest for both `token` and `secret` (e.g., SHA-256 via Web Crypto `crypto.subtle.digest`) and compare the resulting ArrayBuffers/Uint8Arrays byte-by-byte in constant time (or use a timing-safe helper like `timingSafeEqual`) before returning `new Response('Unauthorized', { status: 401 })` when they do not match; update the check in the `job-maintenance.ts` handler where `token` and `secret` are compared.packages/worker/src/mcp/generated-ui-api.ts-145-149 (1)
145-149:⚠️ Potential issue | 🟡 MinorUnhandled failures from
resolveSavedAppSource.This call is outside the existing
try/catch(lines 151–169), so any failure in source resolution (e.g., transient repo/RPC error) will bubble up as an unhandled router 500 instead of a structured JSON error. Previously, parameter parsing errors were reported via a 400 JSON response; consider wrapping this call similarly so clients receive a useful payload, and distinguish between "can't load source" (likely 502/500) and "bad params" (400).♻️ Suggested handling
- const resolvedApp = await resolveSavedAppSource({ - env, - baseUrl: getAppBaseUrl({ env, requestUrl: request.url }), - artifact: app, - }) + let resolvedApp + try { + resolvedApp = await resolveSavedAppSource({ + env, + baseUrl: getAppBaseUrl({ env, requestUrl: request.url }), + artifact: app, + }) + } catch (error) { + return jsonResponse( + { + error: + error instanceof Error + ? error.message + : 'Unable to load saved UI source.', + }, + 502, + ) + }🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@packages/worker/src/mcp/generated-ui-api.ts` around lines 145 - 149, The call to resolveSavedAppSource is unwrapped and can throw, causing unhandled 500s; wrap the resolveSavedAppSource({... baseUrl: getAppBaseUrl({ env, requestUrl: request.url }), artifact: app }) invocation in a try/catch like the existing block handling parameter parsing, and return a structured JSON error instead of letting it bubble: catch parameter/validation errors and respond with a 400 JSON payload, and for source loading or RPC/repo failures respond with a 502/500 JSON payload (include a clear message and any safe error details), matching the existing response shape used elsewhere in this file.packages/worker/src/mcp/capabilities/meta/meta-save-skill.ts-190-197 (1)
190-197:⚠️ Potential issue | 🟡 MinorNon-atomic update sequence across
entity_sourcesandskills.If
updateEntitySourcesucceeds but the subsequent vector upsert fails and triggers the rollback branch below, the entity source'spublishedCommit/indexedCommitwill have moved forward while the skill data has been reverted (or deleted). On next save/re-run,syncArtifactSourceSnapshotmay think the source is already current and skip re-publishing. Consider moving theupdateEntitySourcecall after the successfulupsertSkillVector, or otherwise making the commit pointer update conditional on the full flow succeeding.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@packages/worker/src/mcp/capabilities/meta/meta-save-skill.ts` around lines 190 - 197, The update to entity_sources is happening before the vector upsert, creating a non-atomic state if upsertSkillVector fails; move the updateEntitySource call (the publishedCommit/indexedCommit write) to after a successful upsertSkillVector, or wrap both operations in a single transactional/conditional flow so the commit pointer is only advanced when upsertSkillVector completes successfully (ensure references to updateEntitySource, upsertSkillVector and the rollback branch/syncArtifactSourceSnapshot logic are updated accordingly).packages/worker/src/mcp/capabilities/repo/repo-run-checks.ts-17-19 (1)
17-19:⚠️ Potential issue | 🟡 Minor
readOnly: truelikely misrepresentsrepo_run_checks.Running the Worker-native validation pipeline persists check-run metadata (tree_hash, results, checked_at) on the session — that's precisely what
repo_get_check_statusreads back. This capability also performs meaningful compute, so marking itreadOnly: trueandidempotent: truewill cause callers/routers that rely on capability flags (e.g., for caching, retry semantics, or gating in higher-level agents) to treat it as a free-side-effect query.Suggest
readOnly: false;idempotentis debatable but at minimum the read-only claim is incorrect.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@packages/worker/src/mcp/capabilities/repo/repo-run-checks.ts` around lines 17 - 19, The capability metadata for repo_run_checks incorrectly declares readOnly: true; update the capability descriptor for repo_run_checks to set readOnly: false (leave idempotent as-is or discuss separately) so callers/routers don't treat this operation as a no-side-effect query—locate the capability object named repo_run_checks in repo-run-checks.ts and change the readOnly flag to false while keeping destructive and idempotent settings consistent with current semantics.packages/worker/src/mcp/capabilities/repo/repo-open-session.ts-30-41 (1)
30-41:⚠️ Potential issue | 🟡 MinorExisting-session resume ignores
args.source_idmismatch.If the caller passes
source_id=Awithconversation_id=Cbut the active session bound to that conversation was opened forsource_id=B, this silently returns the session forBand dropsAon the floor. That's easy to misuse when a client rotates between sources within the same conversation. Consider either (a) validatingexistingSession.source_id === args.source_idand erroring on mismatch, or (b) explicitly documenting in the capability description that conversation-bound resume takes precedence oversource_id.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@packages/worker/src/mcp/capabilities/repo/repo-open-session.ts` around lines 30 - 41, The current resume logic returns an existing session found by getActiveRepoSessionByConversation without checking args.source_id, allowing a session opened for a different source to be returned; update the logic in repo-open-session.ts to compare existingSession.source_id with args.source_id and, if they differ, reject (throw an error or return an explicit RPC error) instead of silently returning the mismatched session; ensure the check occurs before calling repoSessionRpc(...).getSessionInfo and include existingSession, args.source_id, and conversation id in the error context for clarity.packages/worker/src/mcp/capabilities/apps/ui-save-app.ts-164-171 (1)
164-171:⚠️ Potential issue | 🟡 Minor
ensureEntitySourceruns before the update-path existence check.When
isUpdateis true and the artifact doesn't belong to the user, we throw at line 293, butensureEntitySourcehas already created a newentity_sourcesrow (or bound a fresh one). MovingensureEntitySourceinto each branch — after the existence check for updates, and where the new appId is committed for creates — avoids leaving orphan entity-source rows on auth/lookup failures.Also applies to: 286-294
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@packages/worker/src/mcp/capabilities/apps/ui-save-app.ts` around lines 164 - 171, ensureEntitySource is being called unconditionally before verifying update permissions, which creates/binds an entity_sources row even when the subsequent existence/auth checks fail; move the ensureEntitySource call into each branch so it runs only after checks pass: for the update path (when isUpdate is true) first verify the artifact/app belongs to the user (the same logic that currently throws in the update branch), then call ensureEntitySource({ db: ctx.env.APP_DB, env: ctx.env, userId: user.userId, entityKind: 'app', entityId: appId, sourceRoot: '/' }); for the create path call ensureEntitySource only after the new appId has been committed/created; update references to ensureEntitySource in the save flow (including the code around the current isUpdate check and the create-commit path) so orphan entity_sources rows are not created on auth/lookup failure.packages/worker/src/mcp/capabilities/repo/repo-tree.ts-85-97 (1)
85-97:⚠️ Potential issue | 🟡 Minor
total_bytesaggregates sizes for directories and symlinks.
totals.total_bytes += current.sizeruns unconditionally, so any non-zerosizereported for a directory or symlink by the RPC contributes to the total. If directories ever report an aggregate size (e.g. recursive total or a link target size) this will double-count or produce a misleading byte total. Safer to only accumulate file sizes.🛠️ Proposed fix
totals.max_depth = Math.max(totals.max_depth, depth) - totals.total_bytes += current.size switch (current.type) { case 'file': totals.files += 1 + totals.total_bytes += current.size break case 'directory': totals.directories += 1 break case 'symlink': totals.symlinks += 1 break }🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@packages/worker/src/mcp/capabilities/repo/repo-tree.ts` around lines 85 - 97, The totals.total_bytes is being increased for every node type, causing directories or symlinks with non-zero size to be incorrectly counted; modify the aggregation so totals.total_bytes += current.size only runs for file nodes (i.e., inside the 'file' case or guarded by if (current.type === 'file')), leaving totals.files, totals.directories, and totals.symlinks updates unchanged (reference the totals and current variables and the switch on current.type in repo-tree.ts).packages/worker/src/mcp/ui-artifacts-repo.ts-27-28 (1)
27-28:⚠️ Potential issue | 🟡 MinorInconsistent null handling for
clientCode: type saysstring, but writes allownulland reads coerce to empty string.The
UiArtifactRow.clientCodetype is declared asstring(non-nullable), yet INSERT (line 28) and UPDATE (line 126) both writeclientCode ?? null, allowing null values into the database. On read,mapRowcoercesnullback to an empty string (String(row['client_code'] ?? '')), creating an asymmetry: savingnull→ storingnull→ reading back as''.This violates the type contract and silently hides missing values. Consumers expecting a string per the type will see empty strings instead of being able to distinguish between "not set" and "legitimately empty".
Align by either:
- Changing
UiArtifactRow.clientCodetostring | nulland havingmapRowreturnnull(so callers can fall through to repo-backed source resolution), or- Keeping it non-nullable and rejecting
nullwrites in INSERT/UPDATE.Also applies to lines 126 and 178.
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@packages/worker/src/mcp/ui-artifacts-repo.ts` around lines 27 - 28, UiArtifactRow.clientCode should be nullable to match the DB writes and reads: change the UiArtifactRow type to clientCode: string | null, update mapRow in ui-artifacts-repo.ts to return row['client_code'] ?? null (do not coerce to an empty string), and keep the INSERT/UPDATE code (the places currently using clientCode ?? null) as-is so nulls are persisted; also audit callers of UiArtifactRow to handle string | null accordingly (particularly code paths that rely on repo-backed source resolution).packages/worker/src/repo/source-sync.ts-12-20 (1)
12-20:⚠️ Potential issue | 🟡 Minor
canSyncArtifactSourceshould also verifyAPP_DB.The guard returns true as long as
ARTIFACTSandREPO_SESSIONare bound, but line 31 unconditionally readsinput.env.APP_DB. In any environment missing the DB binding, this will throw rather than fall back to thereturn nullsemantics the rest of the guard promises.♻️ Proposed fix
function canSyncArtifactSource(env: Env) { return ( typeof (env as Env & { ARTIFACTS?: unknown }).ARTIFACTS === 'object' && (env as Env & { ARTIFACTS?: unknown }).ARTIFACTS != null && typeof (env as Env & { REPO_SESSION?: unknown }).REPO_SESSION === 'object' && - (env as Env & { REPO_SESSION?: unknown }).REPO_SESSION != null + (env as Env & { REPO_SESSION?: unknown }).REPO_SESSION != null && + (env as Env & { APP_DB?: unknown }).APP_DB != null ) }🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@packages/worker/src/repo/source-sync.ts` around lines 12 - 20, canSyncArtifactSource currently only checks ARTIFACTS and REPO_SESSION but code later reads input.env.APP_DB unconditionally; update canSyncArtifactSource to also verify APP_DB is present and non-null (e.g. check typeof (env as Env & { APP_DB?: unknown }).APP_DB === 'object' && (env as Env & { APP_DB?: unknown }).APP_DB != null) so callers can safely access input.env.APP_DB without throwing; use the same casting pattern used for ARTIFACTS and REPO_SESSION to locate the change in the function.packages/worker/src/mcp/capabilities/repo/repo-publish-session.ts-49-56 (1)
49-56:⚠️ Potential issue | 🟡 MinorRemove unnecessary empty-string fallback; RPC contract guarantees non-null.
When
status === 'ok', the type definition guaranteespublishedCommitis astring(non-nullable). The?? ''is dead code sinceRepoSessionPublishResultexplicitly types it as non-null for the ok variant. Either remove the coalescing operator or add an explicit assertion if the contract changes in future.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@packages/worker/src/mcp/capabilities/repo/repo-publish-session.ts` around lines 49 - 56, Remove the unnecessary fallback when constructing the success return: the RPC contract's ok variant (RepoSessionPublishResult) guarantees publishedCommit is a non-null string, so drop the "?? ''" coalescing on result.publishedCommit and return result.publishedCommit directly (or, if you prefer to be explicit, use a non-null assertion on result.publishedCommit). Update the return in the branch checking result.status === 'ok' to use result.publishedCommit as-is and ensure types still satisfy the function's return shape.packages/worker/src/repo/checks.ts-146-150 (1)
146-150:⚠️ Potential issue | 🟡 MinorLint check unconditionally reports "passed".
The
'lint'result is hard-codedok: truewith a placeholder message. As-is, operators looking at check output can't distinguish "lint is intentionally unimplemented" from "lint actually passed", and the aggregateokon Line 162 will never reflect lint failures. At minimum, add aTODO/issue reference; ideally flip the message to something likeLint check is not yet implemented.so it's obvious in any UI.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@packages/worker/src/repo/checks.ts` around lines 146 - 150, The lint result currently pushed via results.push({ kind: 'lint', ok: true, ... }) is a hard-coded placeholder; change it to clearly indicate the check is unimplemented by setting ok: false (or at least not true) and updating the message to something like "Lint check is not yet implemented." also add a TODO or issue reference string in the message so operators can tell this is intentional; locate the results.push call for kind 'lint' and update the ok and message fields so the aggregate ok calculation that consumes results reflects the unimplemented state.packages/worker/src/repo/entity-sources.ts-97-133 (1)
97-133:⚠️ Potential issue | 🟡 MinorThe
.bind()arity is correct, but add test coverage for partial updates.The dynamic SET clause construction properly matches the bind arguments:
valuescontains exactly one entry per SET placeholder (from the conditionaladd()calls plus the always-runupdated_at), and the two WHERE placeholders (idanduser_id) are correctly appended. However, this function currently has no test coverage. Since production code exercises partial updates with different field combinations (see usages insource-service.ts,meta-save-skill.ts, andrepo-session-do.ts), add at least one test verifying a single-field partial update to prevent regressions if fields are added or removed later.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@packages/worker/src/repo/entity-sources.ts` around lines 97 - 133, Add a unit/integration test that exercises updateEntitySource with a partial update (e.g., only publishedCommit or only manifestPath) to ensure the dynamic SET clause and bind ordering are correct; specifically, call updateEntitySource(...) with a single optional field changed and verify the DB row is updated (result true and the specific column changed while others remain unchanged). Target the updateEntitySource function in packages/worker/src/repo/entity-sources.ts and mirror realistic call patterns similar to usages in source-service.ts, meta-save-skill.ts, or repo-session-do.ts; seed a test row, invoke updateEntitySource with just one field, then read the row back to assert only that field and updated_at changed.packages/worker/src/repo/source-templates.ts-50-51 (1)
50-51:⚠️ Potential issue | 🟡 MinorDuplicating
job.nameas bothtitleanddescription.
description: input.job.nameduplicates the title.manifestBaseSchema.descriptionrequiresmin(1), and consumers use both fields independently for search/ranking, so they'll double-weight the same string. Consider usingscheduleSummaryor a stored job description field.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@packages/worker/src/repo/source-templates.ts` around lines 50 - 51, The manifest template currently sets both title and description to the same value (title: input.job.name, description: input.job.name), which duplicates content and skews search/ranking; update the description field in the source-templates generation (where manifestBaseSchema is populated) to use a different source such as input.job.scheduleSummary or a dedicated job description field instead of input.job.name, ensuring manifestBaseSchema.description still meets min(1) validation and consumers receive distinct title and description values.packages/worker/src/mcp/capabilities/unified-search.ts-388-396 (1)
388-396:⚠️ Potential issue | 🟡 MinorJob
descriptionis hardcoded torow.name— drop it or source from a real field.Both
buildJobEmbedDoc(line 391) androwToJobHit(line 516) emitdescription: row.name. As a result:
- The embed doc and hit artifacts contain the job name twice, so lexical/embed ranking is effectively double-weighted on name.
- The rendered hit shows identical title and description to users/LLM callers.
If
JobViewhas no real description field, consider usingscheduleSummaryas the description fallback (it's already the most meaningful secondary text) or leaving it as an empty string instead of duplicating the name.♻️ Proposed change
function buildJobEmbedDoc(row: JobView): string { return buildJobEmbedText({ name: row.name, - description: row.name, + description: row.scheduleSummary ?? '', scheduleSummary: row.scheduleSummary, sourceId: row.sourceId, publishedCommit: row.publishedCommit, }) } @@ return { type: 'job', jobId: row.id, domain: 'jobs', title: row.name, - description: row.name, + description: row.scheduleSummary ?? '', scheduleSummary: row.scheduleSummary,Also applies to: 505-526
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@packages/worker/src/mcp/capabilities/unified-search.ts` around lines 388 - 396, The description field is incorrectly duplicated from the job name in buildJobEmbedDoc and rowToJobHit; update both places (functions buildJobEmbedDoc and rowToJobHit dealing with JobView) to stop setting description: row.name and instead set description to a meaningful fallback such as row.scheduleSummary (or '' if scheduleSummary is missing). Ensure the embed/doc and hit artifacts use description: row.scheduleSummary ?? '' (or equivalent) so the title and description are not identical.packages/worker/src/repo/repo-session-do.ts-541-586 (1)
541-586:⚠️ Potential issue | 🟡 Minor
truncatedsignal can be wrong.
searchInTextbreaks early on each file whenmatches.length >= input.maxMatches, but never returns a signal that it hit the cap. At the caller:truncated: totalMatches > maxMatches
- If a single file had exactly
maxMatches + kreal matches but we stopped atmaxMatches, and no other file matched,totalMatches === maxMatches→truncated = falseeven though we dropped hits.- Conversely,
totalMatches > maxMatchesonly reflects aggregate volume, not the per-file truncation that actually occurred.Make
searchInTextreturn a booleantruncated(or the pre-cap count), and OR-reduce it across files.🛠️ Suggested shape
- const matches = searchInText({ ... }) + const { matches, truncated: fileTruncated } = searchInText({ ... }) if (matches.length === 0) continue + anyFileTruncated ||= fileTruncated ... totalMatches += matches.length } ... - truncated: totalMatches > maxMatches, + truncated: anyFileTruncated,🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@packages/worker/src/repo/repo-session-do.ts` around lines 541 - 586, search truncation is currently unnoticed because searchInText stops per-file at maxMatches but doesn't report that, so change searchInText to return both the matches array and a boolean truncated flag (or the raw pre-cap count); then in repo-session-do.ts where you call searchInText (the block that populates matches, matchMap, totalMatches and uses maxMatches), capture the per-file truncated value and OR-reduce it into an overall truncated variable (e.g., let truncated = false; truncated ||= fileTruncated), and use that final truncated in the returned object instead of computing truncated: totalMatches > maxMatches; keep existing names (searchInText, matches, maxMatches, totalMatches, matchMap) so the change is easy to locate.packages/worker/src/repo/artifacts.ts-125-131 (1)
125-131:⚠️ Potential issue | 🟡 Minor
buildAuthenticatedArtifactsRemotesilently corrupts non-https remotes.
input.remote.slice('https://'.length)unconditionally strips 8 characters. Ifremoteishttp://...or lacks a scheme (e.g., a relative/local test URL), the result is a malformed URL likehttps://x:TOKEN@ttp://host/...and downstream git clone/push operations will fail in confusing ways.🛠️ Proposed hardening
export function buildAuthenticatedArtifactsRemote(input: { remote: string token: string }) { const tokenSecret = parseArtifactTokenSecret(input.token) - return `https://x:${tokenSecret}@${input.remote.slice('https://'.length)}` + if (!input.remote.startsWith('https://')) { + throw new Error(`Artifact remote must use https://, got: ${input.remote}`) + } + return `https://x:${tokenSecret}@${input.remote.slice('https://'.length)}` }🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@packages/worker/src/repo/artifacts.ts` around lines 125 - 131, The function buildAuthenticatedArtifactsRemote currently blindly slices off 8 characters which corrupts non-https remotes; instead, parse the token with parseArtifactTokenSecret(token) into tokenSecret and construct the authenticated remote robustly (use the URL parser or conditional logic) so you only strip/insert the auth part when the scheme is present: for a proper URL create a URL object from input.remote, set url.username = 'x' and url.password = tokenSecret and return url.toString(), and for schemeless strings prepend "https://x:tokenSecret@" or otherwise handle "http://" by replacing its scheme with https and inserting the credentials. Ensure you update the return from buildAuthenticatedArtifactsRemote and keep parseArtifactTokenSecret referenced.packages/worker/src/mcp/capabilities/repo/repo-shared.ts-267-276 (1)
267-276:⚠️ Potential issue | 🟡 Minor
repoRunChecksOutputSchemaintentionally omitsrunId,treeHash, andcheckedAt.The
RepoSessionCheckRuntype returns these fields, but the handler explicitly strips them and only returnsok,results, andmanifest. If these fields are valuable to callers, consider adding them to the schema and handler return value.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@packages/worker/src/mcp/capabilities/repo/repo-shared.ts` around lines 267 - 276, repoRunChecksOutputSchema currently only validates ok, results, and manifest while RepoSessionCheckRun includes runId, treeHash, and checkedAt; update repoRunChecksOutputSchema to include runId (string), treeHash (string), and checkedAt (string or Date ISO) with the same types used by RepoSessionCheckRun and adjust the handler that returns the check run to include these fields in its response so the runtime return shape matches the schema and RepoSessionCheckRun type.
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro
Run ID: 7bae31ae-94a8-47cb-b8ce-48e474caefbf
⛔ Files ignored due to path filters (1)
package-lock.jsonis excluded by!**/package-lock.json
📒 Files selected for processing (69)
packages/shared/src/chat.tspackages/worker/migrations/0023-entity-sources.sqlpackages/worker/migrations/0024-repo-sessions-and-source-columns.sqlpackages/worker/package.jsonpackages/worker/src/agent-turn/tools.tspackages/worker/src/app/handlers/saved-ui.tspackages/worker/src/index.tspackages/worker/src/job-maintenance.tspackages/worker/src/jobs/job-reindex.tspackages/worker/src/jobs/repo.tspackages/worker/src/jobs/service.node.test.tspackages/worker/src/jobs/service.tspackages/worker/src/jobs/types.tspackages/worker/src/mcp/app-runner.tspackages/worker/src/mcp/capabilities/apps/ui-get-app.tspackages/worker/src/mcp/capabilities/apps/ui-load-app-source.tspackages/worker/src/mcp/capabilities/apps/ui-save-app.node.test.tspackages/worker/src/mcp/capabilities/apps/ui-save-app.tspackages/worker/src/mcp/capabilities/builtin-domains.tspackages/worker/src/mcp/capabilities/domain-metadata.tspackages/worker/src/mcp/capabilities/jobs/shared.tspackages/worker/src/mcp/capabilities/meta/meta-save-skill.tspackages/worker/src/mcp/capabilities/repo/domain.tspackages/worker/src/mcp/capabilities/repo/index.tspackages/worker/src/mcp/capabilities/repo/repo-apply-patch.tspackages/worker/src/mcp/capabilities/repo/repo-discard-session.tspackages/worker/src/mcp/capabilities/repo/repo-get-check-status.tspackages/worker/src/mcp/capabilities/repo/repo-get-session.tspackages/worker/src/mcp/capabilities/repo/repo-open-session.tspackages/worker/src/mcp/capabilities/repo/repo-publish-session.tspackages/worker/src/mcp/capabilities/repo/repo-read-file.tspackages/worker/src/mcp/capabilities/repo/repo-rebase-session.tspackages/worker/src/mcp/capabilities/repo/repo-run-checks.tspackages/worker/src/mcp/capabilities/repo/repo-search.tspackages/worker/src/mcp/capabilities/repo/repo-shared.tspackages/worker/src/mcp/capabilities/repo/repo-tree.tspackages/worker/src/mcp/capabilities/repo/repo-write-file.tspackages/worker/src/mcp/capabilities/unified-search.tspackages/worker/src/mcp/context.node.test.tspackages/worker/src/mcp/context.tspackages/worker/src/mcp/executor.tspackages/worker/src/mcp/generated-ui-api.tspackages/worker/src/mcp/jobs-embed.tspackages/worker/src/mcp/jobs-vectorize.tspackages/worker/src/mcp/observability.workers.test.tspackages/worker/src/mcp/run-codemode-registry.tspackages/worker/src/mcp/skills/mcp-skills-repo.tspackages/worker/src/mcp/skills/mcp-skills-types.tspackages/worker/src/mcp/skills/run-saved-skill.tspackages/worker/src/mcp/tools/search-format.tspackages/worker/src/mcp/tools/search.node.test.tspackages/worker/src/mcp/tools/search.tspackages/worker/src/mcp/ui-artifacts-repo.tspackages/worker/src/mcp/ui-artifacts-types.tspackages/worker/src/oauth-handlers-skill-runner.workers.test.tspackages/worker/src/oauth-handlers.tspackages/worker/src/repo/app-source.tspackages/worker/src/repo/artifacts.tspackages/worker/src/repo/checks.tspackages/worker/src/repo/entity-sources.tspackages/worker/src/repo/manifest.tspackages/worker/src/repo/repo-session-do.tspackages/worker/src/repo/repo-sessions.tspackages/worker/src/repo/source-service.tspackages/worker/src/repo/source-sync.tspackages/worker/src/repo/source-templates.tspackages/worker/src/repo/types.tspackages/worker/worker-configuration.d.tspackages/worker/wrangler.jsonc
| const texts = batch.map(({ job }) => | ||
| buildJobEmbedText({ | ||
| name: job.name, | ||
| description: job.name, | ||
| scheduleSummary: job.scheduleSummary, | ||
| sourceId: job.sourceId, | ||
| publishedCommit: job.publishedCommit, | ||
| }), | ||
| ) |
There was a problem hiding this comment.
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash
# Confirm buildJobEmbedText's expected inputs and how the primary write path calls it.
fd -t f 'jobs-embed.ts' | xargs -I{} sed -n '1,120p' {}
rg -nP -C3 '\bbuildJobEmbedText\s*\('Repository: kentcdodds/kody
Length of output: 3613
🏁 Script executed:
#!/bin/bash
# Check the Job type definition to see if description field exists
rg -nP 'type Job|interface Job' --max-count=10
rg -nP '\bjob\.description\b'
# Also check if any entity has a description field
rg -nP '\bdescription:' packages/worker/src/jobs/types.ts -A2 -B2Repository: kentcdodds/kody
Length of output: 3031
description is wired to job.name — embedding input is wrong.
description: job.name duplicates the name into the description slot. Job objects have a description field that should be used instead. This corrupts embeddings across all reindex, create, and update paths, causing search quality to degrade silently.
🐛 Proposed fix
const texts = batch.map(({ job }) =>
buildJobEmbedText({
name: job.name,
- description: job.name,
+ description: job.description,
scheduleSummary: job.scheduleSummary,
sourceId: job.sourceId,
publishedCommit: job.publishedCommit,
}),
)Note: The same bug exists in packages/worker/src/jobs/service.ts (lines 196 and 315) and packages/worker/src/mcp/capabilities/unified-search.ts (line 391). All call sites pass entity name to description instead of the actual description field.
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@packages/worker/src/jobs/job-reindex.ts` around lines 45 - 53, The embedding
input incorrectly sets description: job.name when calling buildJobEmbedText in
job-reindex.ts; change it to description: job.description so the actual job
description is embedded (also search for other call sites like the
buildJobEmbedText calls in jobs/service.ts and
mcp/capabilities/unified-search.ts where description is similarly passed as
job.name and fix them to use job.description). Ensure you update every
invocation of buildJobEmbedText that currently uses job.name for the description
parameter so embeddings reflect the true description field.
Co-authored-by: Kent C. Dodds <me+github@kentcdodds.com>
Co-authored-by: Kent C. Dodds <me+github@kentcdodds.com> Co-authored-by: Kent C. Dodds <me+github@kentcdodds.com>
Move repo session search, RPC typing, and shared session result types into focused modules so future changes stay local and easier to review. Made-with: Cursor
Move repo session path, tree, and session-info shaping into a focused helper module so the session DO stays centered on orchestration. Made-with: Cursor
There was a problem hiding this comment.
Actionable comments posted: 5
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@packages/worker/src/jobs/service.ts`:
- Around line 253-272: The updateJob flow allows swapping sourceId without
validation because ensureEntitySource is only called when shape.sourceId &&
!existing.sourceId; change this to run whenever the source is changing
(shape.sourceId && shape.sourceId !== existing.sourceId) or alternatively reject
updates that attempt to change an existing source; update the conditional that
calls ensureEntitySource (referencing ensureEntitySource, shape.sourceId,
existing.sourceId) so ensuredSource is produced for changes and ensure
updated.sourceId (and subsequent calls to syncArtifactSourceSnapshot and
upsertJobVector) only use a vetted ensuredSource or the original
existing.sourceId.
In `@packages/worker/src/repo/repo-session-do.ts`:
- Around line 636-640: runRepoChecks is being called with
repoSessionWorkspacePrefix which ignores individual source.source_root and
causes snapshot keys to be misaligned; resolve source.source_root against the
workspace prefix (the same way manifestPath is resolved earlier) to compute an
effective sourceRoot and pass that to runRepoChecks instead of
repoSessionWorkspacePrefix. Locate where runRepoChecks is invoked (the call
using manifestPath and repoSessionWorkspacePrefix) and replace the sourceRoot
argument with the resolved path (join/resolve workspace prefix +
source.source_root, handle leading/trailing slashes) so
snapshot.read('package.json') and other reads use the correct relative keys.
- Around line 283-352: In openSession, enforce ownership in both branches: after
retrieving sessionRow from getRepoSessionById, validate sessionRow.user_id ===
input.userId and throw if not before calling
resolveSessionRepo/ensureArtifactRepoRemote/initialize/toRepoSessionInfoResult;
and in the create branch after getEntitySourceById but before
forking/insertRepoSession, validate source.user_id === input.userId and throw if
not to prevent cross-user forks; update the checks adjacent to
getRepoSessionById and getEntitySourceById in the openSession flow.
In `@packages/worker/src/repo/repo-session-search.ts`:
- Around line 155-189: The current code uses maxMatches (from
normalizeSearchLimit(input.limit)) as a per-file cap because maxMatches is
computed once and passed into searchInText for every file; to make limit a
global budget, compute remaining = normalizeSearchLimit(input.limit) before the
files loop, pass remaining (not maxMatches) into searchInText for each file,
after adding matches decrement remaining by the number of matches actually
appended and increment totalMatches, set truncated true if result.truncated OR
if remaining reaches 0, and break out of the outer files loop when remaining <=
0; adjust uses of maxMatches/remaining and references to searchInText, matchMap,
truncated, totalMatches, and the files loop accordingly (alternatively, if you
meant a per-file limit, rename maxMatches to perFileLimit and document it).
In `@packages/worker/src/repo/repo-session-tree.ts`:
- Around line 46-50: The check in resolveRepoWorkspacePath uses
trimmed.startsWith(workspacePrefix) which misclassifies sibling paths (e.g.,
'/session-backup'); update the guard to treat an exact match (trimmed ===
workspacePrefix) or a proper directory prefix
(trimmed.startsWith(workspacePrefix + '/')) so only true children are returned
unchanged; otherwise normalize by returning
`${workspacePrefix}/${trimmed.replace(/^\/+/, '')}`. Ensure you modify the
conditional in resolveRepoWorkspacePath to use these two precise checks instead
of the loose startsWith.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro
Run ID: 73a1f3e1-b709-422a-b264-d7a760ffe1a8
📒 Files selected for processing (6)
packages/worker/src/jobs/service.tspackages/worker/src/repo/repo-session-do.tspackages/worker/src/repo/repo-session-rpc.tspackages/worker/src/repo/repo-session-search.tspackages/worker/src/repo/repo-session-tree.tspackages/worker/src/repo/types.ts
There was a problem hiding this comment.
Cursor Bugbot has reviewed your changes and found 2 potential issues.
There are 4 total unresolved issues (including 2 from previous reviews).
Autofix Details
Bugbot Autofix prepared fixes for both issues found in the latest run.
- ✅ Fixed: Deterministic session ID enables concurrent request race condition
- Generated a unique repo session ID per resolve call to prevent shared sessions and discard races.
- ✅ Fixed: Job
storageIdthrow on null breaks existing rows without it- Fallback to a derived job storage ID when rows lack storage_id to avoid crashes on legacy data.
Preview (ad414e84ac)
diff --git a/package-lock.json b/package-lock.json
--- a/package-lock.json
+++ b/package-lock.json
@@ -678,7 +678,6 @@
"resolved": "https://registry.npmjs.org/@cloudflare/shell/-/shell-0.3.2.tgz",
"integrity": "sha512-2UYlhp308IxKWutGcdOJ8+QTR2J1sck108JJ0AEDYG/aNaapKSCrm0Rvd3EFAfqAPtzsqtFqDOKU/4LtpCF8dw==",
"license": "MIT",
- "peer": true,
"dependencies": {
"@cloudflare/codemode": "^0.3.4",
"isomorphic-git": "^1.37.5"
@@ -1212,6 +1211,38 @@
"url": "https://github.com/sponsors/colinhacks"
}
},
+ "node_modules/@cloudflare/worker-bundler": {
+ "version": "0.1.1",
+ "resolved": "https://registry.npmjs.org/@cloudflare/worker-bundler/-/worker-bundler-0.1.1.tgz",
+ "integrity": "sha512-pwEgMFk7RORJZD7sg0GgnW47zATPAFCJocfiiGCEJ6qjeD+QFENwnp495Nudim1LkPkt0EsERQykNfV/c3gaaA==",
+ "license": "MIT",
+ "dependencies": {
+ "@typescript/vfs": "^1.6.4",
+ "es-module-lexer": "^2.0.0",
+ "esbuild-wasm": "0.28.0",
+ "resolve.exports": "^2.0.3",
+ "semver": "^7.7.4",
+ "smol-toml": "^1.6.1",
+ "sucrase": "^3.35.1",
+ "typescript": "^6.0.2"
+ },
+ "engines": {
+ "node": ">=22"
+ }
+ },
+ "node_modules/@cloudflare/worker-bundler/node_modules/typescript": {
+ "version": "6.0.2",
+ "resolved": "https://registry.npmjs.org/typescript/-/typescript-6.0.2.tgz",
+ "integrity": "sha512-bGdAIrZ0wiGDo5l8c++HWtbaNCWTS4UTv7RaTH/ThVIgjkveJt83m74bBHMJkuCbslY8ixgLBVZJIOiQlQTjfQ==",
+ "license": "Apache-2.0",
+ "bin": {
+ "tsc": "bin/tsc",
+ "tsserver": "bin/tsserver"
+ },
+ "engines": {
+ "node": ">=14.17"
+ }
+ },
"node_modules/@cloudflare/workerd-darwin-64": {
"version": "1.20260415.1",
"resolved": "https://registry.npmjs.org/@cloudflare/workerd-darwin-64/-/workerd-darwin-64-1.20260415.1.tgz",
@@ -1412,6 +1443,7 @@
"cpu": [
"ppc64"
],
+ "dev": true,
"license": "MIT",
"optional": true,
"os": [
@@ -1428,6 +1460,7 @@
"cpu": [
"arm"
],
+ "dev": true,
"license": "MIT",
"optional": true,
"os": [
@@ -1444,6 +1477,7 @@
"cpu": [
"arm64"
],
+ "dev": true,
"license": "MIT",
"optional": true,
"os": [
@@ -1460,6 +1494,7 @@
"cpu": [
"x64"
],
+ "dev": true,
"license": "MIT",
"optional": true,
"os": [
@@ -1493,6 +1528,7 @@
"cpu": [
"x64"
],
+ "dev": true,
"license": "MIT",
"optional": true,
"os": [
@@ -1509,6 +1545,7 @@
"cpu": [
"arm64"
],
+ "dev": true,
"license": "MIT",
"optional": true,
"os": [
@@ -1525,6 +1562,7 @@
"cpu": [
"x64"
],
+ "dev": true,
"license": "MIT",
"optional": true,
"os": [
@@ -1541,6 +1579,7 @@
"cpu": [
"arm"
],
+ "dev": true,
"license": "MIT",
"optional": true,
"os": [
@@ -1557,6 +1596,7 @@
"cpu": [
"arm64"
],
+ "dev": true,
"license": "MIT",
"optional": true,
"os": [
@@ -1573,6 +1613,7 @@
"cpu": [
"ia32"
],
+ "dev": true,
"license": "MIT",
"optional": true,
"os": [
@@ -1589,6 +1630,7 @@
"cpu": [
"loong64"
],
+ "dev": true,
"license": "MIT",
"optional": true,
"os": [
@@ -1605,6 +1647,7 @@
"cpu": [
"mips64el"
],
+ "dev": true,
"license": "MIT",
"optional": true,
"os": [
@@ -1621,6 +1664,7 @@
"cpu": [
"ppc64"
],
+ "dev": true,
"license": "MIT",
"optional": true,
"os": [
@@ -1637,6 +1681,7 @@
"cpu": [
"riscv64"
],
+ "dev": true,
"license": "MIT",
"optional": true,
"os": [
@@ -1653,6 +1698,7 @@
"cpu": [
"s390x"
],
+ "dev": true,
"license": "MIT",
"optional": true,
"os": [
@@ -1669,6 +1715,7 @@
"cpu": [
"x64"
],
+ "dev": true,
"license": "MIT",
"optional": true,
"os": [
@@ -1685,6 +1732,7 @@
"cpu": [
"arm64"
],
+ "dev": true,
"license": "MIT",
"optional": true,
"os": [
@@ -1701,6 +1749,7 @@
"cpu": [
"x64"
],
+ "dev": true,
"license": "MIT",
"optional": true,
"os": [
@@ -1717,6 +1766,7 @@
"cpu": [
"arm64"
],
+ "dev": true,
"license": "MIT",
"optional": true,
"os": [
@@ -1733,6 +1783,7 @@
"cpu": [
"x64"
],
+ "dev": true,
"license": "MIT",
"optional": true,
"os": [
@@ -1749,6 +1800,7 @@
"cpu": [
"arm64"
],
+ "dev": true,
"license": "MIT",
"optional": true,
"os": [
@@ -1765,6 +1817,7 @@
"cpu": [
"x64"
],
+ "dev": true,
"license": "MIT",
"optional": true,
"os": [
@@ -1781,6 +1834,7 @@
"cpu": [
"arm64"
],
+ "dev": true,
"license": "MIT",
"optional": true,
"os": [
@@ -1797,6 +1851,7 @@
"cpu": [
"ia32"
],
+ "dev": true,
"license": "MIT",
"optional": true,
"os": [
@@ -1813,6 +1868,7 @@
"cpu": [
"x64"
],
+ "dev": true,
"license": "MIT",
"optional": true,
"os": [
@@ -4088,9 +4144,6 @@
"arm64"
],
"dev": true,
- "libc": [
- "glibc"
- ],
"license": "MIT",
"optional": true,
"os": [
@@ -4108,9 +4161,6 @@
"arm64"
],
"dev": true,
- "libc": [
- "musl"
- ],
"license": "MIT",
"optional": true,
"os": [
@@ -4128,9 +4178,6 @@
"ppc64"
],
"dev": true,
- "libc": [
- "glibc"
- ],
"license": "MIT",
"optional": true,
"os": [
@@ -4148,9 +4195,6 @@
"riscv64"
],
"dev": true,
- "libc": [
- "glibc"
- ],
"license": "MIT",
"optional": true,
"os": [
@@ -4168,9 +4212,6 @@
"riscv64"
],
"dev": true,
- "libc": [
- "musl"
- ],
"license": "MIT",
"optional": true,
"os": [
@@ -4188,9 +4229,6 @@
"s390x"
],
"dev": true,
- "libc": [
- "glibc"
- ],
"license": "MIT",
"optional": true,
"os": [
@@ -4208,9 +4246,6 @@
"x64"
],
"dev": true,
- "libc": [
- "glibc"
- ],
"license": "MIT",
"optional": true,
"os": [
@@ -4228,9 +4263,6 @@
"x64"
],
"dev": true,
- "libc": [
- "musl"
- ],
"license": "MIT",
"optional": true,
"os": [
@@ -5577,9 +5609,6 @@
"cpu": [
"arm64"
],
- "libc": [
- "glibc"
- ],
"license": "MIT",
"optional": true,
"os": [
@@ -5596,9 +5625,6 @@
"cpu": [
"arm64"
],
- "libc": [
- "musl"
- ],
"license": "MIT",
"optional": true,
"os": [
@@ -5615,9 +5641,6 @@
"cpu": [
"ppc64"
],
- "libc": [
- "glibc"
- ],
"license": "MIT",
"optional": true,
"os": [
@@ -5634,9 +5657,6 @@
"cpu": [
"s390x"
],
- "libc": [
- "glibc"
- ],
"license": "MIT",
"optional": true,
"os": [
@@ -5653,9 +5673,6 @@
"cpu": [
"x64"
],
- "libc": [
- "glibc"
- ],
"license": "MIT",
"optional": true,
"os": [
@@ -5672,9 +5689,6 @@
"cpu": [
"x64"
],
- "libc": [
- "musl"
- ],
"license": "MIT",
"optional": true,
"os": [
@@ -5904,9 +5918,6 @@
"arm"
],
"dev": true,
- "libc": [
- "glibc"
- ],
"license": "MIT",
"optional": true,
"os": [
@@ -5921,9 +5932,6 @@
"arm"
],
"dev": true,
- "libc": [
- "musl"
- ],
"license": "MIT",
"optional": true,
"os": [
@@ -5937,9 +5945,6 @@
"cpu": [
"arm64"
],
- "libc": [
- "glibc"
- ],
"license": "MIT",
"optional": true,
"os": [
@@ -5954,9 +5959,6 @@
"arm64"
],
"dev": true,
- "libc": [
- "musl"
- ],
"license": "MIT",
"optional": true,
"os": [
@@ -5971,9 +5973,6 @@
"loong64"
],
"dev": true,
- "libc": [
- "glibc"
- ],
"license": "MIT",
"optional": true,
"os": [
@@ -5988,9 +5987,6 @@
"loong64"
],
"dev": true,
- "libc": [
- "musl"
- ],
"license": "MIT",
"optional": true,
"os": [
@@ -6005,9 +6001,6 @@
"ppc64"
],
"dev": true,
- "libc": [
- "glibc"
- ],
"license": "MIT",
"optional": true,
"os": [
@@ -6022,9 +6015,6 @@
"ppc64"
],
"dev": true,
- "libc": [
- "musl"
- ],
"license": "MIT",
"optional": true,
"os": [
@@ -6039,9 +6029,6 @@
"riscv64"
],
"dev": true,
- "libc": [
- "glibc"
- ],
"license": "MIT",
"optional": true,
"os": [
@@ -6056,9 +6043,6 @@
"riscv64"
],
"dev": true,
- "libc": [
- "musl"
- ],
"license": "MIT",
"optional": true,
"os": [
@@ -6073,9 +6057,6 @@
"s390x"
],
"dev": true,
- "libc": [
- "glibc"
- ],
"license": "MIT",
"optional": true,
"os": [
@@ -6089,9 +6070,6 @@
"cpu": [
"x64"
],
- "libc": [
- "glibc"
- ],
"license": "MIT",
"optional": true,
"os": [
@@ -6106,9 +6084,6 @@
"x64"
],
"dev": true,
- "libc": [
- "musl"
- ],
"license": "MIT",
"optional": true,
"os": [
@@ -8086,7 +8061,6 @@
},
"node_modules/@typescript/vfs": {
"version": "1.6.4",
- "dev": true,
"license": "MIT",
"dependencies": {
"debug": "^4.4.3"
@@ -9257,7 +9231,6 @@
"resolved": "https://registry.npmjs.org/abort-controller/-/abort-controller-3.0.0.tgz",
"integrity": "sha512-h8lQ8tacZYnR3vNQTgibj+tODHI5/+l06Au2Pcriv/Gmet0eaj4TwWH41sO9wnHDiQsEj19q0drzdWdeAHtweg==",
"license": "MIT",
- "peer": true,
"dependencies": {
"event-target-shim": "^5.0.0"
},
@@ -9585,6 +9558,12 @@
"node": ">=14"
}
},
+ "node_modules/any-promise": {
+ "version": "1.3.0",
+ "resolved": "https://registry.npmjs.org/any-promise/-/any-promise-1.3.0.tgz",
+ "integrity": "sha512-7UvmKalWRt1wgjL1RrGxoSJW/0QZFIegpeGvZG9kjp8vrRu55XTHbwnqq2GpXm9uLbcuhxm3IqX9OB4MZR1b2A==",
+ "license": "MIT"
+ },
"node_modules/anymatch": {
"version": "3.1.3",
"dev": true,
@@ -9791,8 +9770,7 @@
"version": "1.4.1",
"resolved": "https://registry.npmjs.org/async-lock/-/async-lock-1.4.1.tgz",
"integrity": "sha512-Az2ZTpuytrtqENulXwO3GGv1Bztugx6TT37NIo7imr/Qo0gsYiGtSdBa2B6fsXhTpVZDNfu1Qn3pk531e3q+nQ==",
- "license": "MIT",
- "peer": true
+ "license": "MIT"
},
"node_modules/asynckit": {
"version": "0.4.0",
@@ -10250,8 +10228,7 @@
"version": "2.0.1",
"resolved": "https://registry.npmjs.org/clean-git-ref/-/clean-git-ref-2.0.1.tgz",
"integrity": "sha512-bLSptAy2P0s6hU4PzuIMKmMJJSE6gLXGH1cntDu7bWJUksvuM+7ReOK61mozULErYvP6a15rnYl0zFDef+pyPw==",
- "license": "Apache-2.0",
- "peer": true
+ "license": "Apache-2.0"
},
"node_modules/cli-cursor": {
"version": "3.1.0",
@@ -10650,7 +10627,6 @@
"resolved": "https://registry.npmjs.org/crc-32/-/crc-32-1.2.2.tgz",
"integrity": "sha512-ROmzCKrTnOwybPcJApAA6WBWij23HVfGVNKqqrZpuyZOHqK2CwHSvpGuyt/UNNvaIjEd8X5IFGp4Mh+Ie1IHJQ==",
"license": "Apache-2.0",
- "peer": true,
"bin": {
"crc32": "bin/crc32.njs"
},
@@ -11278,7 +11254,6 @@
"resolved": "https://registry.npmjs.org/decompress-response/-/decompress-response-6.0.0.tgz",
"integrity": "sha512-aW35yZM6Bb/4oJlZncMH2LCoZtJXTRxES17vE3hoRiowU2kWHaJKFkSBDnDR+cm9J+9QhXmREyIfv0pji9ejCQ==",
"license": "MIT",
- "peer": true,
"dependencies": {
"mimic-response": "^3.1.0"
},
@@ -11443,8 +11418,7 @@
"version": "0.0.3",
"resolved": "https://registry.npmjs.org/diff3/-/diff3-0.0.3.tgz",
"integrity": "sha512-iSq8ngPOt0K53A6eVr4d5Kn6GNrM2nQZtC740pzIriHtn4pOQ2lyzEXQMBeVcWERN0ye7fhBsk9PbLLQOnUx/g==",
- "license": "MIT",
- "peer": true
+ "license": "MIT"
},
"node_modules/dns-equal": {
"version": "1.0.0",
@@ -11826,7 +11800,6 @@
},
"node_modules/es-module-lexer": {
"version": "2.0.0",
- "dev": true,
"license": "MIT"
},
"node_modules/es-object-atoms": {
@@ -11884,7 +11857,7 @@
"version": "0.27.7",
"resolved": "https://registry.npmjs.org/esbuild/-/esbuild-0.27.7.tgz",
"integrity": "sha512-IxpibTjyVnmrIQo5aqNpCgoACA/dTKLTlhMHihVHhdkxKyPO1uBBthumT0rdHmcsk9uMonIWS0m4FljWzILh3w==",
- "devOptional": true,
+ "dev": true,
"hasInstallScript": true,
"license": "MIT",
"bin": {
@@ -11922,6 +11895,18 @@
"@esbuild/win32-x64": "0.27.7"
}
},
+ "node_modules/esbuild-wasm": {
+ "version": "0.28.0",
+ "resolved": "https://registry.npmjs.org/esbuild-wasm/-/esbuild-wasm-0.28.0.tgz",
+ "integrity": "sha512-5TRVKExcEmeMkccIZMzUq+Az6X2RoMAJyfl6SMMO1dMVhmvt0I2mx7gAb6zYi42n4d1ETcatFXazGKzA+aW7fg==",
+ "license": "MIT",
+ "bin": {
+ "esbuild": "bin/esbuild"
+ },
+ "engines": {
+ "node": ">=18"
+ }
+ },
"node_modules/esbuild/node_modules/@esbuild/darwin-arm64": {
"version": "0.27.7",
"resolved": "https://registry.npmjs.org/@esbuild/darwin-arm64/-/darwin-arm64-0.27.7.tgz",
@@ -11929,6 +11914,7 @@
"cpu": [
"arm64"
],
+ "dev": true,
"license": "MIT",
"optional": true,
"os": [
@@ -12415,7 +12401,6 @@
"resolved": "https://registry.npmjs.org/event-target-shim/-/event-target-shim-5.0.1.tgz",
"integrity": "sha512-i/2XbnSz/uxRCU6+NdVJgKWDTM427+MqYbkQzD321DuCQJUqOuJKIA0IM2+W2xtYHdKOmZ4dR6fExsd4SXL+WQ==",
"license": "MIT",
- "peer": true,
"engines": {
"node": ">=6"
}
@@ -12430,7 +12415,6 @@
"resolved": "https://registry.npmjs.org/events/-/events-3.3.0.tgz",
"integrity": "sha512-mQw+2fkQbALzQ7V0MY0IqdnXNOeTtP4r0lN9z7AAawCXgqea7bDii20AYrIBrFd/Hx0M2Ocz6S111CaFkUcb0Q==",
"license": "MIT",
- "peer": true,
"engines": {
"node": ">=0.8.x"
}
@@ -12893,6 +12877,7 @@
"version": "2.3.3",
"resolved": "https://registry.npmjs.org/fsevents/-/fsevents-2.3.3.tgz",
"integrity": "sha512-5xoDfX+fL7faATnagmWPpbFtwh/R77WmMMqqHGS65C3vvB0YHrgF+B1YmZ3441tMj5n63k0212XNoJwzlhffQw==",
+ "dev": true,
"hasInstallScript": true,
"license": "MIT",
"optional": true,
@@ -14086,7 +14071,6 @@
"resolved": "https://registry.npmjs.org/isomorphic-git/-/isomorphic-git-1.37.5.tgz",
"integrity": "sha512-wek54c5uFvd3WsxewLWt6h0GXKWQh0P8rRXns9bN1rHNjcgCb3+0lmyAsP594NeTtQFeCJQVS9b0kjbkD1l5qg==",
"license": "MIT",
- "peer": true,
"dependencies": {
"async-lock": "^1.4.1",
"clean-git-ref": "^2.0.1",
@@ -14126,7 +14110,6 @@
}
],
"license": "MIT",
- "peer": true,
"dependencies": {
"base64-js": "^1.3.1",
"ieee754": "^1.2.1"
@@ -14137,7 +14120,6 @@
"resolved": "https://registry.npmjs.org/ignore/-/ignore-5.3.2.tgz",
"integrity": "sha512-hsBTNUqQTDwkWtcdYI2i06Y/nUBEsNEDJKjWdigLvegy8kDuJAS8uRlpkkcQpyEXL0Z/pjDy5HBmMjRCJ2gq+g==",
"license": "MIT",
- "peer": true,
"engines": {
"node": ">= 4"
}
@@ -14147,7 +14129,6 @@
"resolved": "https://registry.npmjs.org/readable-stream/-/readable-stream-4.7.0.tgz",
"integrity": "sha512-oIGGmcpTLwPga8Bn6/Z75SVaH1z5dUut2ibSyAMVhmUggWpmDn2dapB0n7f8nwaSiRtepAsfJyfXIO5DCVAODg==",
"license": "MIT",
- "peer": true,
"dependencies": {
"abort-controller": "^3.0.0",
"buffer": "^6.0.3",
@@ -14207,7 +14188,7 @@
},
"node_modules/jiti": {
"version": "2.6.1",
- "devOptional": true,
+ "dev": true,
"license": "MIT",
"bin": {
"jiti": "lib/jiti-cli.mjs"
@@ -14531,6 +14512,7 @@
"cpu": [
"arm64"
],
+ "dev": true,
"license": "MPL-2.0",
"optional": true,
"os": [
@@ -14551,6 +14533,7 @@
"cpu": [
"arm64"
],
+ "dev": true,
"license": "MPL-2.0",
"optional": true,
"os": [
@@ -14571,6 +14554,7 @@
"cpu": [
"x64"
],
+ "dev": true,
"license": "MPL-2.0",
"optional": true,
"os": [
@@ -14591,6 +14575,7 @@
"cpu": [
"x64"
],
+ "dev": true,
"license": "MPL-2.0",
"optional": true,
"os": [
@@ -14611,6 +14596,7 @@
"cpu": [
"arm"
],
+ "dev": true,
"license": "MPL-2.0",
"optional": true,
"os": [
@@ -14631,6 +14617,7 @@
"cpu": [
"arm64"
],
+ "dev": true,
"license": "MPL-2.0",
"optional": true,
"os": [
@@ -14651,6 +14638,7 @@
"cpu": [
"arm64"
],
+ "dev": true,
"license": "MPL-2.0",
"optional": true,
"os": [
@@ -14671,6 +14659,7 @@
"cpu": [
"x64"
],
+ "dev": true,
"license": "MPL-2.0",
"optional": true,
"os": [
@@ -14691,6 +14680,7 @@
"cpu": [
"x64"
],
+ "dev": true,
"license": "MPL-2.0",
"optional": true,
"os": [
@@ -14711,6 +14701,7 @@
"cpu": [
"arm64"
],
+ "dev": true,
"license": "MPL-2.0",
"optional": true,
"os": [
@@ -14731,6 +14722,7 @@
"cpu": [
"x64"
],
+ "dev": true,
"license": "MPL-2.0",
"optional": true,
"os": [
@@ -16056,7 +16048,6 @@
"resolved": "https://registry.npmjs.org/mimic-response/-/mimic-response-3.1.0.tgz",
"integrity": "sha512-z0yWI+4FDrrweS8Zmt4Ej5HdJmky15+L2e6Wgn3+iK5fWzb6T3fhNFq2+MeTRb064c6Wr4N/wv0DzQTjNzHNGQ==",
"license": "MIT",
- "peer": true,
"engines": {
"node": ">=10"
},
... diff truncated: showing 800 of 8639 linesYou can send follow-ups to the cloud agent here.
Co-authored-by: Kent C. Dodds <me+github@kentcdodds.com>
| publishedCommit?: string | null | ||
| }) { | ||
| return typeof input.sourceId === 'string' && input.sourceId.length > 0 | ||
| } |
There was a problem hiding this comment.
Unused publishedCommit parameter in hasRepoBackedJobSource
Low Severity
hasRepoBackedJobSource declares publishedCommit in its input type but never reads it. The function only checks sourceId. This is misleading since callers pass objects containing both fields, giving the impression that publishedCommit participates in the decision.
Reviewed by Cursor Bugbot for commit 7d96125. Configure here.
Co-authored-by: Kent C. Dodds <me+github@kentcdodds.com>
Co-authored-by: Kent C. Dodds <me+github@kentcdodds.com>
|
|
||
| const savedAppSourceCache = new Map<string, ResolvedSavedAppSource>() | ||
| const savedAppSourceCacheLimit = 100 | ||
|
|
There was a problem hiding this comment.
Module-level Map cache leaks memory across Worker requests
Low Severity
savedAppSourceCache is a module-level Map with a 100-entry LRU cap. In Cloudflare Workers, module-scoped state persists across requests within the same isolate. While the cap prevents unbounded growth, the cached ResolvedSavedAppSource objects contain full clientCode and serverCode strings that can be large, and the cache is never cleared between requests. Under steady traffic to many different apps, this memory pressure persists for the lifetime of the isolate.
Reviewed by Cursor Bugbot for commit e544738. Configure here.
There was a problem hiding this comment.
Cursor Bugbot has reviewed your changes and found 2 potential issues.
There are 4 total unresolved issues (including 2 from previous reviews).
Bugbot Autofix is ON, but it could not run because the branch was deleted or merged before autofix could start.
Reviewed by Cursor Bugbot for commit ddb50b0. Configure here.
| .catch(() => { | ||
| // Best effort only; ensure new sessions pick up the latest commit. | ||
| }) | ||
| session = await sessionClient.openSession(openSessionInput) |
There was a problem hiding this comment.
Stale session reused when discard silently fails
Medium Severity
In runRepoBackedJob, when base_commit differs from published_commit, discardSession is called with .catch(() => {}), swallowing all errors. If the discard fails (e.g., D1 delete fails), the subsequent openSession finds the old session row still in the database and reuses it with the stale base_commit, causing the job to execute against outdated code instead of the latest published commit.
Reviewed by Cursor Bugbot for commit ddb50b0. Configure here.
| /\bmodule\.exports\b/.test(code) || | ||
| /\bexports\.[A-Za-z_$][\w$]*/.test(code) | ||
| ) | ||
| } |
There was a problem hiding this comment.
Module-style regex falsely matches content inside template literals
Low Severity
hasModuleStyleCodemodeEntrypoint uses /^\s*export\s+/m which, due to the multiline flag, matches any line starting with export — including lines inside template literals or multi-line strings. A legitimate async function snippet containing a template literal with an export-prefixed line would be incorrectly rejected as a module-style entrypoint, blocking execution of valid repo-backed job code.
Reviewed by Cursor Bugbot for commit ddb50b0. Configure here.
2 participants
Summary
repo_*capability surface for opening sessions, reading/searching files, summarizing trees, applying transactional edits, running Worker-native checks, and rebase/publish lifecycle operationsgojs/esbuild.wasmpackage bugTesting
npm run typechecknpx vitest run packages/worker/src/mcp/context.node.test.ts packages/worker/src/jobs/service.node.test.ts packages/worker/src/oauth-handlers-skill-runner.workers.test.tsnpx vitest run packages/worker/src/jobs/service.node.test.ts packages/worker/src/mcp/run-codemode-registry.node.test.ts packages/worker/src/mcp/tools/search.node.test.ts packages/worker/src/mcp/capabilities/meta/meta-list-capabilities.node.test.ts packages/worker/src/mcp/capabilities/build-capability-registry.workers.test.tsnpx vitest run packages/worker/src/mcp/tools/search.node.test.ts packages/worker/src/mcp/capabilities/unified-search.workers.test.tsnpx vitest run packages/worker/src/mcp/capabilities/apps/ui-save-app.node.test.ts packages/worker/src/mcp/observability.workers.test.ts packages/worker/src/mcp/capabilities/apps/saved-app-capabilities.node.test.ts packages/worker/src/app/saved-ui-hosted-html.node.test.tsnpx vitest run packages/worker/src/storage-runner.workers.test.ts packages/worker/src/mcp/capabilities/build-capability-registry.workers.test.tsnpm run preview:e2egit pushruns exercised the branch through the repo’s pre-push test/e2e hooks on each stable checkpointNotes
@cloudflare/worker-bundlerpackage still has a broken eager import path in this environment (gojsthroughesbuild.wasm), so the repo checks path intentionally lazy-loads it to avoid breaking unrelated imports while the migration continuesSummary by CodeRabbit
New Features
Chores