Skip to content

Open-source Grok Bot alternative for named AI teammates. Local-first desktop, web, and mobile clients with per-bot tools, instructions, and memory. Independent fork of T3 Code.

phase 1

Julius Marminge6954d1b91c34
diff --git a/.plans/16-pr89-review-remediation-phases.md b/.plans/16-pr89-review-remediation-phases.md
index 513727eda..4adafa656 100644
--- a/.plans/16-pr89-review-remediation-phases.md
+++ b/.plans/16-pr89-review-remediation-phases.md
@@ -10,6 +10,7 @@
- Invalid/false-positive: 3 (excluded)
- Duplicate reposts: collapsed
- Unique actionable findings after filtering: 58
+ - Post-rewrite validity audit: 5 additional stale items marked invalid, leaving 53 actionable (`34 valid` + `19 partially-valid`)
## Phase 0: Canonical Triage Lock
diff --git a/.plans/16a-pr89-phase0-canonical-triage.json b/.plans/16a-pr89-phase0-canonical-triage.json
index 761146cf8..8c6a93cd6 100644
--- a/.plans/16a-pr89-phase0-canonical-triage.json
+++ b/.plans/16a-pr89-phase0-canonical-triage.json
@@ -10,11 +10,11 @@
"phase0_generated_at": "2026-02-26T02:49:21.091145Z",
"summary": {
"total_threads": 185,
- "actionable_canonical": 58,
+ "actionable_canonical": 53,
"actionable_duplicates": 24,
"filtered_outdated": 94,
"filtered_resolved": 6,
- "filtered_invalid": 3
+ "filtered_invalid": 8
},
"canonical_items": [
{
@@ -379,24 +379,6 @@
"macroscopeapp"
]
},
- {
- "canonical_id": "C021",
- "area": "Other",
- "title": "Shared mutable default metadata object causes stale eventId",
- "severity": "Medium",
- "path": "apps/server/src/orchestration/decider.ts",
- "line": 27,
- "thread_ids": [
- "PRRT_kwDORLtfbc5wkPaA"
- ],
- "thread_indexes": [
- 128
- ],
- "duplicate_count": 0,
- "authors": [
- "cursor"
- ]
- },
{
"canonical_id": "C022",
"area": "Edge-case parsing/platform behavior",
@@ -451,24 +433,6 @@
"macroscopeapp"
]
},
- {
- "canonical_id": "C025",
- "area": "Checkpointing correctness",
- "title": "Duplicated checkpoint ref computation across two files",
- "severity": "Medium",
- "path": "apps/server/src/wsServer.ts",
- "line": 128,
- "thread_ids": [
- "PRRT_kwDORLtfbc5wvwag"
- ],
- "thread_indexes": [
- 148
- ],
- "duplicate_count": 0,
- "authors": [
- "cursor"
- ]
- },
{
"canonical_id": "C026",
"area": "WebSocket robustness",
@@ -561,24 +525,6 @@
"macroscopeapp"
]
},
- {
- "canonical_id": "C031",
- "area": "Event ordering and state consistency",
- "title": "Revert uses wrong turn count from positional inference",
- "severity": "Medium",
- "path": "apps/web/src/session-logic.ts",
- "line": 127,
- "thread_ids": [
- "PRRT_kwDORLtfbc5v9SCp"
- ],
- "thread_indexes": [
- 41
- ],
- "duplicate_count": 0,
- "authors": [
- "cursor"
- ]
- },
{
"canonical_id": "C032",
"area": "Other",
@@ -657,24 +603,6 @@
"cursor"
]
},
- {
- "canonical_id": "C036",
- "area": "Checkpointing correctness",
- "title": "Duplicate `checkpointRefForThreadTurn` function in two production files",
- "severity": "Low",
- "path": "apps/server/src/checkpointing/Layers/CheckpointStore.ts",
- "line": 284,
- "thread_ids": [
- "PRRT_kwDORLtfbc5wiqFX"
- ],
- "thread_indexes": [
- 115
- ],
- "duplicate_count": 0,
- "authors": [
- "cursor"
- ]
- },
{
"canonical_id": "C037",
"area": "Other",
@@ -1003,24 +931,6 @@
"macroscopeapp"
]
},
- {
- "canonical_id": "C055",
- "area": "Checkpointing correctness",
- "title": "Duplicate `checkpointRefForThreadTurn` function across files",
- "severity": "Low",
- "path": "apps/server/src/wsServer.ts",
- "line": 128,
- "thread_ids": [
- "PRRT_kwDORLtfbc5wkPaG"
- ],
- "thread_indexes": [
- 129
- ],
- "duplicate_count": 0,
- "authors": [
- "cursor"
- ]
- },
{
"canonical_id": "C056",
"area": "Other",
@@ -1684,9 +1594,8 @@
"title": "Revert uses wrong turn count from positional inference",
"severity": "Medium",
"has_false_positive_signal": false,
- "canonical_id": "C031",
- "phase0_status": "actionable_canonical",
- "phase0_reason": "Canonical actionable item."
+ "phase0_status": "filtered_invalid",
+ "phase0_reason": "Invalid after post-rewrite audit: Revert now uses explicit checkpointTurnCount first; positional fallback is non-primary."
},
{
"index": 42,
@@ -2740,9 +2649,8 @@
"title": "Duplicate `checkpointRefForThreadTurn` function in two production files",
"severity": "Low",
"has_false_positive_signal": false,
- "canonical_id": "C036",
- "phase0_status": "actionable_canonical",
- "phase0_reason": "Canonical actionable item."
+ "phase0_status": "filtered_invalid",
+ "phase0_reason": "Invalid after post-rewrite audit: No longer duplicated; single production source via Refs.ts."
},
{
"index": 116,
@@ -2929,9 +2837,8 @@
"title": "Shared mutable default metadata object causes stale eventId",
"severity": "Medium",
"has_false_positive_signal": false,
- "canonical_id": "C021",
- "phase0_status": "actionable_canonical",
- "phase0_reason": "Canonical actionable item."
+ "phase0_status": "filtered_invalid",
+ "phase0_reason": "Invalid after post-rewrite audit: Stale-eventId claim no longer applies; eventId is regenerated per event."
},
{
"index": 129,
@@ -2944,9 +2851,8 @@
"title": "Duplicate `checkpointRefForThreadTurn` function across files",
"severity": "Low",
"has_false_positive_signal": false,
- "canonical_id": "C055",
- "phase0_status": "actionable_canonical",
- "phase0_reason": "Canonical actionable item."
+ "phase0_status": "filtered_invalid",
+ "phase0_reason": "Invalid after post-rewrite audit: No longer duplicated; helper is centralized."
},
{
"index": 130,
@@ -3226,9 +3132,8 @@
"title": "Duplicated checkpoint ref computation across two files",
"severity": "Medium",
"has_false_positive_signal": false,
- "canonical_id": "C025",
- "phase0_status": "actionable_canonical",
- "phase0_reason": "Canonical actionable item."
+ "phase0_status": "filtered_invalid",
+ "phase0_reason": "Invalid after post-rewrite audit: No longer duplicated; checkpoint ref helper now centralized."
},
{
"index": 149,
diff --git a/.plans/16a-pr89-phase0-canonical-triage.md b/.plans/16a-pr89-phase0-canonical-triage.md
index e59ef61dc..6bbe9d385 100644
--- a/.plans/16a-pr89-phase0-canonical-triage.md
+++ b/.plans/16a-pr89-phase0-canonical-triage.md
@@ -7,20 +7,18 @@
## Summary
- Total threads: 185
-- Actionable canonical: 58
+- Actionable canonical: 53
- Actionable duplicates: 24
- Filtered outdated: 94
- Filtered resolved: 6
-- Filtered invalid: 3
+- Filtered invalid: 8
## Canonical Actionable Checklist
-### Other (17)
+### Other (16)
- [ ] C011 [Medium] Template literal `${cause}` throws if `cause` is a `Symbol`. Consider using `String(cause)` instead. (apps/server/src/orchestration/Errors.ts:117)
- Threads: PRRT_kwDORLtfbc5wBFv4
-- [ ] C021 [Medium] Shared mutable default metadata object causes stale eventId (apps/server/src/orchestration/decider.ts:27)
- - Threads: PRRT_kwDORLtfbc5wkPaA
- [ ] C024 [Medium] In `executeUnprepared`, preparing outside an effect and not closing the statement can cause uncaught errors and leaks. Suggest wrapping with `Effect.acquireUseRelease`: prepare in acquire (error-safe) and `finalize` in release. (apps/server/src/persistence/NodeSqliteClient.ts:145)
- Threads: PRRT_kwDORLtfbc5whxX9
- [ ] C028 [Medium] Branch sync dispatches both server and stale local update (apps/web/src/components/BranchToolbar.tsx:102)
@@ -54,15 +52,15 @@
### Runtime resilience and failure handling (10)
-- [ ] C002 [High] A dispatch error in `processEvent` will terminate the `Effect.forever` loop, permanently halting event ingestion. Consider adding error recovery (e.g., `Effect.catchAll` with logging) around `processEvent` so failures don't kill the fiber. (apps/server/src/orchestration/Layers/ProviderRuntimeIngestion.ts:333) (+5 dup)
+- [x] C002 [High] A dispatch error in `processEvent` will terminate the `Effect.forever` loop, permanently halting event ingestion. Consider adding error recovery (e.g., `Effect.catchAll` with logging) around `processEvent` so failures don't kill the fiber. (apps/server/src/orchestration/Layers/ProviderRuntimeIngestion.ts:333) (+5 dup)
- Threads: PRRT_kwDORLtfbc5wj4cH, PRRT_kwDORLtfbc5wnWwF, PRRT_kwDORLtfbc5wyTaP, PRRT_kwDORLtfbc5wzliw, PRRT_kwDORLtfbc5w0_g3, PRRT_kwDORLtfbc5w1HGT
-- [ ] C012 [Medium] Forked revert dispatch risks read model inconsistency (apps/server/src/orchestration/Layers/CheckpointReactor.ts:542) (+4 dup)
+- [x] C012 [Medium] Forked revert dispatch risks read model inconsistency (apps/server/src/orchestration/Layers/CheckpointReactor.ts:542) (+4 dup)
- Threads: PRRT_kwDORLtfbc5whszW, PRRT_kwDORLtfbc5wyTaS, PRRT_kwDORLtfbc5wzli0, PRRT_kwDORLtfbc5w0_g4, PRRT_kwDORLtfbc5w1HGX
- [ ] C019 [Medium] ProviderRuntimeIngestion processes events for wrong thread on race (apps/server/src/orchestration/Layers/ProviderRuntimeIngestion.ts:178)
- Threads: PRRT_kwDORLtfbc5wkPaL
- [ ] C020 [Medium] On `message.completed`, the message ID is added to the set and `thread.message.assistant.complete` is dispatched. On `turn.completed`, the same set is iterated and `thread.message.assistant.complete` is dispatched again for each ID—including already-completed ones. Consider removing message IDs from the set after dispatching on `message.completed`, or filtering out already-completed IDs before the `turn.completed` loop. (apps/server/src/orchestration/Layers/ProviderRuntimeIngestion.ts:266)
- Threads: PRRT_kwDORLtfbc5w1GPr
-- [ ] C027 [Medium] WS message handler can cause unhandled promise rejection (apps/server/src/wsServer.ts:545) (+1 dup)
+- [x] C027 [Medium] WS message handler can cause unhandled promise rejection (apps/server/src/wsServer.ts:545) (+1 dup)
- Threads: PRRT_kwDORLtfbc5wyTaW, PRRT_kwDORLtfbc5wzli3
- [ ] C042 [Low] Duplicated `resolveThreadWorkspaceCwd` across three files (apps/server/src/orchestration/Layers/CheckpointReactor.ts:62)
- Threads: PRRT_kwDORLtfbc5wzli2
@@ -96,26 +94,7 @@
- [ ] C040 [Low] `upsertKeybindingRule` has a race condition: concurrent calls read the same file state, then the last write overwrites earlier changes. Consider wrapping the read-modify-write sequence with `Effect.Semaphore` to serialize access. <details> <summary>🚀 Reply "<strong>fix it for me</strong>" or copy this <strong>AI Prompt</strong> for your agent:</summary> (apps/server/src/keybindings.ts:488)
- Threads: PRRT_kwDORLtfbc5wxvIA
-### Checkpointing correctness (8)
-
-- [ ] C008 [Medium] Inconsistent input normalization across CheckpointStore methods (apps/server/src/checkpointing/Layers/CheckpointStore.ts:94) (+3 dup)
- - Threads: PRRT_kwDORLtfbc5widJw, PRRT_kwDORLtfbc5wnWv_, PRRT_kwDORLtfbc5w0_g7, PRRT_kwDORLtfbc5w1C36
-- [ ] C017 [Medium] `REQUIRED_SNAPSHOT_PROJECTORS` includes `pending-approvals` and `thread-turns`, but `getSnapshot` doesn't query their data. If these projectors lag behind, the returned `snapshotSequence` will be lower than what the included data actually reflects, causing clients to replay already-applied events. Consider filtering `REQUIRED_SNAPSHOT_PROJECTORS` to only include projectors whose data is actually fetched in the snapshot. <details> <summary>🚀 Reply "<strong>fix it for me</strong>" or copy this <strong>AI Prompt</strong> for your agent:</summary> (apps/server/src/orchestration/Layers/ProjectionSnapshotQuery.ts:71)
- - Threads: PRRT_kwDORLtfbc5wiLhQ
-- [ ] C025 [Medium] Duplicated checkpoint ref computation across two files (apps/server/src/wsServer.ts:128)
- - Threads: PRRT_kwDORLtfbc5wvwag
-- [ ] C033 [Low] Three error classes defined but never instantiated anywhere (apps/server/src/checkpointing/Errors.ts:51)
- - Threads: PRRT_kwDORLtfbc5wlYgo
-- [ ] C034 [Low] Redundant `CheckpointInvariantError` in `CheckpointServiceError` union type (apps/server/src/checkpointing/Errors.ts:79)
- - Threads: PRRT_kwDORLtfbc5wj5fn
-- [ ] C035 [Low] Redundant error type in CheckpointServiceError union definition (apps/server/src/checkpointing/Errors.ts:79) (+2 dup)
- - Threads: PRRT_kwDORLtfbc5wlYgs, PRRT_kwDORLtfbc5wxsO6, PRRT_kwDORLtfbc5w1C4B
-- [ ] C036 [Low] Duplicate `checkpointRefForThreadTurn` function in two production files (apps/server/src/checkpointing/Layers/CheckpointStore.ts:284)
- - Threads: PRRT_kwDORLtfbc5wiqFX
-- [ ] C055 [Low] Duplicate `checkpointRefForThreadTurn` function across files (apps/server/src/wsServer.ts:128)
- - Threads: PRRT_kwDORLtfbc5wkPaG
-
-### Event ordering and state consistency (6)
+### Event ordering and state consistency (5)
- [ ] C001 [High] Non-atomic event appending can corrupt state on retry. If an error occurs mid-loop (lines 96-102) after some events are persisted but before the receipt is written, the command appears to fail. A retry generates new UUIDs via `crypto.randomUUID()` in the decider, appending duplicate events. Consider wrapping the loop in a transaction or using deterministic event IDs derived from `commandId`. <details> <summary>🚀 Reply "<strong>fix it for me</strong>" or copy this <strong>AI Prompt</strong> for your agent:</summary> (apps/server/src/orchestration/Layers/OrchestrationEngine.ts:96)
- Threads: PRRT_kwDORLtfbc5wzp4T
@@ -127,14 +106,12 @@
- Threads: PRRT_kwDORLtfbc5whxJO
- [ ] C016 [Medium] The in-memory `pendingTurnStartByThreadId` map isn't restored during bootstrap. If the service restarts after processing `thread.turn-start-requested` but before `thread.session-set`, the `userMessageId` and `startedAt` will be lost since bootstrap resumes *after* the committed sequence. Consider persisting this pending state or processing these two events atomically. <details> <summary>🚀 Reply "<strong>fix it for me</strong>" or copy this <strong>AI Prompt</strong> for your agent:</summary> (apps/server/src/orchestration/Layers/ProjectionPipeline.ts:490)
- Threads: PRRT_kwDORLtfbc5wxvH8
-- [ ] C031 [Medium] Revert uses wrong turn count from positional inference (apps/web/src/session-logic.ts:127)
- - Threads: PRRT_kwDORLtfbc5v9SCp
### WebSocket robustness (5)
- [ ] C003 [High] Consider attaching a no-op error listener before `socket.write` (e.g., `socket.on('error', () => {})`) to prevent an unhandled `EPIPE`/`ECONNRESET` from crashing the process if the client disconnects mid-handshake. (apps/server/src/wsServer.ts:75)
- Threads: PRRT_kwDORLtfbc5v-cf4
-- [ ] C026 [Medium] Consider adding `.catch(() => {})` after `Effect.runPromise(handleMessage(ws, raw))` to prevent unhandled rejections from crashing the server if `encodeResponse` or setup logic fails. (apps/server/src/wsServer.ts:545)
+- [x] C026 [Medium] Consider adding `.catch(() => {})` after `Effect.runPromise(handleMessage(ws, raw))` to prevent unhandled rejections from crashing the server if `encodeResponse` or setup logic fails. (apps/server/src/wsServer.ts:545)
- Threads: PRRT_kwDORLtfbc5wj4cE
- [ ] C053 [Low] Consider using `socket.end(response)` instead of `socket.write(response)` + `socket.destroy()` to ensure the HTTP error response is fully flushed before closing the connection. (apps/server/src/wsServer.ts:83)
- Threads: PRRT_kwDORLtfbc5v-WPD
@@ -143,6 +120,19 @@
- [ ] C059 [Low] Suggestion: don’t spread `params` into `body`; it can override `_tag` and mishandle non-object values. Keep `_tag` separate and nest `params` under a single key (e.g., `data`), or validate `params` is a plain object. (apps/web/src/wsTransport.ts:59)
- Threads: PRRT_kwDORLtfbc5whtrN
+### Checkpointing correctness (5)
+
+- [ ] C008 [Medium] Inconsistent input normalization across CheckpointStore methods (apps/server/src/checkpointing/Layers/CheckpointStore.ts:94) (+3 dup)
+ - Threads: PRRT_kwDORLtfbc5widJw, PRRT_kwDORLtfbc5wnWv_, PRRT_kwDORLtfbc5w0_g7, PRRT_kwDORLtfbc5w1C36
+- [ ] C017 [Medium] `REQUIRED_SNAPSHOT_PROJECTORS` includes `pending-approvals` and `thread-turns`, but `getSnapshot` doesn't query their data. If these projectors lag behind, the returned `snapshotSequence` will be lower than what the included data actually reflects, causing clients to replay already-applied events. Consider filtering `REQUIRED_SNAPSHOT_PROJECTORS` to only include projectors whose data is actually fetched in the snapshot. <details> <summary>🚀 Reply "<strong>fix it for me</strong>" or copy this <strong>AI Prompt</strong> for your agent:</summary> (apps/server/src/orchestration/Layers/ProjectionSnapshotQuery.ts:71)
+ - Threads: PRRT_kwDORLtfbc5wiLhQ
+- [ ] C033 [Low] Three error classes defined but never instantiated anywhere (apps/server/src/checkpointing/Errors.ts:51)
+ - Threads: PRRT_kwDORLtfbc5wlYgo
+- [ ] C034 [Low] Redundant `CheckpointInvariantError` in `CheckpointServiceError` union type (apps/server/src/checkpointing/Errors.ts:79)
+ - Threads: PRRT_kwDORLtfbc5wj5fn
+- [ ] C035 [Low] Redundant error type in CheckpointServiceError union definition (apps/server/src/checkpointing/Errors.ts:79) (+2 dup)
+ - Threads: PRRT_kwDORLtfbc5wlYgs, PRRT_kwDORLtfbc5wxsO6, PRRT_kwDORLtfbc5w1C4B
+
### Build/runtime portability (2)
- [ ] C006 [Medium] Build script runs TypeScript file with bare node (apps/server/package.json:14)
@@ -161,8 +151,18 @@
- Rationale: Invalid by design: provider runtime events are intentionally surfaced through orchestration domain events (`orchestration.domainEvent`), not a separate `providers.event` WebSocket channel in the current architecture.
- PRRT_kwDORLtfbc5v9An- | apps/web/src/wsNativeApi.ts:165 | push stream, and the callbacks rely on unchecked casts. Please reintroduce a `providers.onEvent` (or equivalent) that subscribes to the `providers.event` channel and decode read-model/domain-event payloads using the shared contracts schemas before invoking callbacks. As per coding guidelines: "Web app must consume provider event streams via WebSocket push on channel `providers.event`" and "Use Zod schemas from `packages/contracts` for shared type contracts covering provider events, WebSocket protocol, and model/session types". <details> <summary>🤖 Prompt for AI Agents</summary>
- Rationale: Invalid by design: provider runtime events are intentionally surfaced through orchestration domain events (`orchestration.domainEvent`), not a separate `providers.event` WebSocket channel in the current architecture.
+- PRRT_kwDORLtfbc5v9SCp | apps/web/src/session-logic.ts:127 | Revert uses wrong turn count from positional inference
+ - Rationale: Invalid after post-rewrite audit: Revert now uses explicit checkpointTurnCount first; positional fallback is non-primary.
- PRRT_kwDORLtfbc5widJx | apps/server/package.json:14 | Build script requires implicit Node.js TypeScript support
- Rationale: Thread contains explicit false-positive determination in replies.
+- PRRT_kwDORLtfbc5wiqFX | apps/server/src/checkpointing/Layers/CheckpointStore.ts:284 | Duplicate `checkpointRefForThreadTurn` function in two production files
+ - Rationale: Invalid after post-rewrite audit: No longer duplicated; single production source via Refs.ts.
+- PRRT_kwDORLtfbc5wkPaA | apps/server/src/orchestration/decider.ts:27 | Shared mutable default metadata object causes stale eventId
+ - Rationale: Invalid after post-rewrite audit: Stale-eventId claim no longer applies; eventId is regenerated per event.
+- PRRT_kwDORLtfbc5wkPaG | apps/server/src/wsServer.ts:128 | Duplicate `checkpointRefForThreadTurn` function across files
+ - Rationale: Invalid after post-rewrite audit: No longer duplicated; helper is centralized.
+- PRRT_kwDORLtfbc5wvwag | apps/server/src/wsServer.ts:128 | Duplicated checkpoint ref computation across two files
+ - Rationale: Invalid after post-rewrite audit: No longer duplicated; checkpoint ref helper now centralized.
## Duplicate Mapping
diff --git a/.plans/16b-pr89-architecture-validity-audit.json b/.plans/16b-pr89-architecture-validity-audit.json
new file mode 100644
index 000000000..ab9407fd7
--- /dev/null
+++ b/.plans/16b-pr89-architecture-validity-audit.json
@@ -0,0 +1,790 @@
+{
+ "source": ".plans/16a-pr89-phase0-canonical-triage.json",
+ "architecture_assumption": "Provider runtime activity is projected and consumed via orchestration.domainEvent; no separate providers.event push channel is expected.",
+ "summary": {
+ "total_canonical_reviewed": 58,
+ "valid": 34,
+ "partially_valid": 19,
+ "invalid": 5
+ },
+ "records": [
+ {
+ "canonical_id": "C001",
+ "verdict": "valid",
+ "rationale": "Append/project/receipt are non-atomic; retry can duplicate events.",
+ "severity": "High",
+ "area": "Event ordering and state consistency",
+ "path": "apps/server/src/orchestration/Layers/OrchestrationEngine.ts",
+ "line": 96,
+ "title": "Non-atomic event appending can corrupt state on retry. If an error occurs mid-loop (lines 96-102) after some events are persisted but before the receipt is written, the command appears to fail. A retry generates new UUIDs via `crypto.randomUUID()` in the decider, appending duplicate events. Consider wrapping the loop in a transaction or using deterministic event IDs derived from `commandId`. <details> <summary>\ud83d\ude80 Reply \"<strong>fix it for me</strong>\" or copy this <strong>AI Prompt</strong> for your agent:</summary>",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wzp4T"
+ ]
+ },
+ {
+ "canonical_id": "C002",
+ "verdict": "valid",
+ "rationale": "Ingestion worker loop can terminate on unhandled processEvent failure.",
+ "severity": "High",
+ "area": "Runtime resilience and failure handling",
+ "path": "apps/server/src/orchestration/Layers/ProviderRuntimeIngestion.ts",
+ "line": 333,
+ "title": "A dispatch error in `processEvent` will terminate the `Effect.forever` loop, permanently halting event ingestion. Consider adding error recovery (e.g., `Effect.catchAll` with logging) around `processEvent` so failures don't kill the fiber.",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wj4cH",
+ "PRRT_kwDORLtfbc5wnWwF",
+ "PRRT_kwDORLtfbc5wyTaP",
+ "PRRT_kwDORLtfbc5wzliw",
+ "PRRT_kwDORLtfbc5w0_g3",
+ "PRRT_kwDORLtfbc5w1HGT"
+ ]
+ },
+ {
+ "canonical_id": "C003",
+ "verdict": "valid",
+ "rationale": "Upgrade reject writes then destroys socket without defensive error listener.",
+ "severity": "High",
+ "area": "WebSocket robustness",
+ "path": "apps/server/src/wsServer.ts",
+ "line": 75,
+ "title": "Consider attaching a no-op error listener before `socket.write` (e.g., `socket.on('error', () => {})`) to prevent an unhandled `EPIPE`/`ECONNRESET` from crashing the process if the client disconnects mid-handshake.",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5v-cf4"
+ ]
+ },
+ {
+ "canonical_id": "C006",
+ "verdict": "partially-valid",
+ "rationale": "Node 24 makes this work, but script still relies on runtime-specific TS execution.",
+ "severity": "Medium",
+ "area": "Build/runtime portability",
+ "path": "apps/server/package.json",
+ "line": 14,
+ "title": "Build script runs TypeScript file with bare node",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wxsO-"
+ ]
+ },
+ {
+ "canonical_id": "C007",
+ "verdict": "partially-valid",
+ "rationale": "Same as C006 (duplicate concern).",
+ "severity": "Medium",
+ "area": "Build/runtime portability",
+ "path": "apps/server/package.json",
+ "line": 14,
+ "title": "Build script uses `node` to run `.ts` file directly",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wiLNS",
+ "PRRT_kwDORLtfbc5wj5fl"
+ ]
+ },
+ {
+ "canonical_id": "C008",
+ "verdict": "partially-valid",
+ "rationale": "Normalization inconsistency remains, but upstream callsites now mostly canonicalize inputs.",
+ "severity": "Medium",
+ "area": "Checkpointing correctness",
+ "path": "apps/server/src/checkpointing/Layers/CheckpointStore.ts",
+ "line": 94,
+ "title": "Inconsistent input normalization across CheckpointStore methods",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5widJw",
+ "PRRT_kwDORLtfbc5wnWv_",
+ "PRRT_kwDORLtfbc5w0_g7",
+ "PRRT_kwDORLtfbc5w1C36"
+ ]
+ },
+ {
+ "canonical_id": "C009",
+ "verdict": "valid",
+ "rationale": "Braced rename parsing still breaks paths like src/{old => new}/file.ts.",
+ "severity": "Medium",
+ "area": "Edge-case parsing/platform behavior",
+ "path": "apps/server/src/git/Layers/GitCore.ts",
+ "line": 41,
+ "title": "Git's braced rename syntax (e.g., `src/{old => new}/file.ts`) isn't handled correctly. The current slice after ` => ` produces invalid paths like `new}/file.ts`. Consider expanding the braces to construct the full destination path. <details> <summary>\ud83d\ude80 Reply \"<strong>fix it for me</strong>\" or copy this <strong>AI Prompt</strong> for your agent:</summary>",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5w1CxT"
+ ]
+ },
+ {
+ "canonical_id": "C010",
+ "verdict": "valid",
+ "rationale": "ENOENT for missing keybindings config still not handled as empty/default.",
+ "severity": "Medium",
+ "area": "Edge-case parsing/platform behavior",
+ "path": "apps/server/src/keybindings.ts",
+ "line": 418,
+ "title": "`loadCustomKeybindingsConfig` fails when the config file doesn't exist, which is expected for new users. Consider catching `ENOENT` and returning an empty array instead. <details> <summary>\ud83d\ude80 Reply \"<strong>fix it for me</strong>\" or copy this <strong>AI Prompt</strong> for your agent:</summary>",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wxvIJ"
+ ]
+ },
+ {
+ "canonical_id": "C011",
+ "verdict": "valid",
+ "rationale": "Template-literal interpolation of Symbol still throws.",
+ "severity": "Medium",
+ "area": "Other",
+ "path": "apps/server/src/orchestration/Errors.ts",
+ "line": 117,
+ "title": "Template literal `${cause}` throws if `cause` is a `Symbol`. Consider using `String(cause)` instead.",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wBFv4"
+ ]
+ },
+ {
+ "canonical_id": "C012",
+ "verdict": "valid",
+ "rationale": "Revert completion dispatch remains forked; state consistency window remains.",
+ "severity": "Medium",
+ "area": "Runtime resilience and failure handling",
+ "path": "apps/server/src/orchestration/Layers/CheckpointReactor.ts",
+ "line": 542,
+ "title": "Forked revert dispatch risks read model inconsistency",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5whszW",
+ "PRRT_kwDORLtfbc5wyTaS",
+ "PRRT_kwDORLtfbc5wzli0",
+ "PRRT_kwDORLtfbc5w0_g4",
+ "PRRT_kwDORLtfbc5w1HGX"
+ ]
+ },
+ {
+ "canonical_id": "C013",
+ "verdict": "valid",
+ "rationale": "Persisted event can outpace in-memory projection on mid-flight failure.",
+ "severity": "Medium",
+ "area": "Event ordering and state consistency",
+ "path": "apps/server/src/orchestration/Layers/OrchestrationEngine.ts",
+ "line": 99,
+ "title": "If `projectionPipeline.projectEvent` fails after `eventStore.append` succeeds, the event is persisted but `readModel` isn't updated, causing desync. Consider updating the in-memory `readModel` immediately after append (before the external projection), so local state stays consistent regardless of downstream failures.",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5whtrM"
+ ]
+ },
+ {
+ "canonical_id": "C014",
+ "verdict": "partially-valid",
+ "rationale": "Broad catch may be intentional for liveness; correctness concern is contextual.",
+ "severity": "Medium",
+ "area": "Event ordering and state consistency",
+ "path": "apps/server/src/orchestration/Layers/OrchestrationEngine.ts",
+ "line": 144,
+ "title": "Engine error handler catches all errors including non-invariant ones",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wkPaJ"
+ ]
+ },
+ {
+ "canonical_id": "C015",
+ "verdict": "partially-valid",
+ "rationale": "Message fallback retention issue is real, but prior FK-violation claim is overstated.",
+ "severity": "Medium",
+ "area": "Event ordering and state consistency",
+ "path": "apps/server/src/orchestration/Layers/ProjectionPipeline.ts",
+ "line": 99,
+ "title": "The gap-filling fallback logic can retain messages from turns that are about to be deleted, causing foreign key violations. Consider removing the fallback logic entirely, or filtering `fallbackUserMessages` and `fallbackAssistantMessages` to only include messages whose `turnId` is in `retainedTurnIds`. <details> <summary>\ud83d\ude80 Reply \"<strong>fix it for me</strong>\" or copy this <strong>AI Prompt</strong> for your agent:</summary>",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5whxJO"
+ ]
+ },
+ {
+ "canonical_id": "C016",
+ "verdict": "valid",
+ "rationale": "Pending turn-start map is in-memory only and not rebuilt on bootstrap.",
+ "severity": "Medium",
+ "area": "Event ordering and state consistency",
+ "path": "apps/server/src/orchestration/Layers/ProjectionPipeline.ts",
+ "line": 490,
+ "title": "The in-memory `pendingTurnStartByThreadId` map isn't restored during bootstrap. If the service restarts after processing `thread.turn-start-requested` but before `thread.session-set`, the `userMessageId` and `startedAt` will be lost since bootstrap resumes *after* the committed sequence. Consider persisting this pending state or processing these two events atomically. <details> <summary>\ud83d\ude80 Reply \"<strong>fix it for me</strong>\" or copy this <strong>AI Prompt</strong> for your agent:</summary>",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wxvH8"
+ ]
+ },
+ {
+ "canonical_id": "C017",
+ "verdict": "partially-valid",
+ "rationale": "Snapshot sequence can under-report due to extra projectors, but replay impact is lower now.",
+ "severity": "Medium",
+ "area": "Checkpointing correctness",
+ "path": "apps/server/src/orchestration/Layers/ProjectionSnapshotQuery.ts",
+ "line": 71,
+ "title": "`REQUIRED_SNAPSHOT_PROJECTORS` includes `pending-approvals` and `thread-turns`, but `getSnapshot` doesn't query their data. If these projectors lag behind, the returned `snapshotSequence` will be lower than what the included data actually reflects, causing clients to replay already-applied events. Consider filtering `REQUIRED_SNAPSHOT_PROJECTORS` to only include projectors whose data is actually fetched in the snapshot. <details> <summary>\ud83d\ude80 Reply \"<strong>fix it for me</strong>\" or copy this <strong>AI Prompt</strong> for your agent:</summary>",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wiLhQ"
+ ]
+ },
+ {
+ "canonical_id": "C018",
+ "verdict": "valid",
+ "rationale": "handledTurnStartKeys still grows without pruning.",
+ "severity": "Medium",
+ "area": "Memory/resource growth",
+ "path": "apps/server/src/orchestration/Layers/ProviderCommandReactor.ts",
+ "line": 84,
+ "title": "Unbounded memory growth in turn start deduplication set",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5whszQ",
+ "PRRT_kwDORLtfbc5wl2A8",
+ "PRRT_kwDORLtfbc5wyTaT",
+ "PRRT_kwDORLtfbc5wzliz",
+ "PRRT_kwDORLtfbc5w0_g-",
+ "PRRT_kwDORLtfbc5w1HGW"
+ ]
+ },
+ {
+ "canonical_id": "C019",
+ "verdict": "partially-valid",
+ "rationale": "SessionId-only routing can misassociate events under races/rebinds.",
+ "severity": "Medium",
+ "area": "Runtime resilience and failure handling",
+ "path": "apps/server/src/orchestration/Layers/ProviderRuntimeIngestion.ts",
+ "line": 178,
+ "title": "ProviderRuntimeIngestion processes events for wrong thread on race",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wkPaL"
+ ]
+ },
+ {
+ "canonical_id": "C020",
+ "verdict": "partially-valid",
+ "rationale": "Duplicate complete dispatch exists; downstream impact often idempotent.",
+ "severity": "Medium",
+ "area": "Runtime resilience and failure handling",
+ "path": "apps/server/src/orchestration/Layers/ProviderRuntimeIngestion.ts",
+ "line": 266,
+ "title": "On `message.completed`, the message ID is added to the set and `thread.message.assistant.complete` is dispatched. On `turn.completed`, the same set is iterated and `thread.message.assistant.complete` is dispatched again for each ID\u2014including already-completed ones. Consider removing message IDs from the set after dispatching on `message.completed`, or filtering out already-completed IDs before the `turn.completed` loop.",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5w1GPr"
+ ]
+ },
+ {
+ "canonical_id": "C021",
+ "verdict": "invalid",
+ "rationale": "Stale-eventId claim no longer applies; eventId is regenerated per event.",
+ "severity": "Medium",
+ "area": "Other",
+ "path": "apps/server/src/orchestration/decider.ts",
+ "line": 27,
+ "title": "Shared mutable default metadata object causes stale eventId",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wkPaA"
+ ]
+ },
+ {
+ "canonical_id": "C022",
+ "verdict": "valid",
+ "rationale": "fish PATH formatting risk still exists in os-jank path recovery.",
+ "severity": "Medium",
+ "area": "Edge-case parsing/platform behavior",
+ "path": "apps/server/src/os-jank.ts",
+ "line": 10,
+ "title": "Fish shell outputs `$PATH` as space-separated, not colon-separated. Consider checking if the shell is fish and using `string join : $PATH` instead, or validating the result contains colons before assigning.",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wkRZM"
+ ]
+ },
+ {
+ "canonical_id": "C023",
+ "verdict": "valid",
+ "rationale": "-ilc shell invocation can pollute captured PATH output.",
+ "severity": "Medium",
+ "area": "Edge-case parsing/platform behavior",
+ "path": "apps/server/src/os-jank.ts",
+ "line": 10,
+ "title": "Using `-il` flags causes the shell to source profile scripts that may print banners or other text, polluting the captured `PATH`. Consider using `-lc` (login only, non-interactive) to reduce unwanted output.",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wj4cM"
+ ]
+ },
+ {
+ "canonical_id": "C024",
+ "verdict": "partially-valid",
+ "rationale": "prepare() still outside managed effect boundary; failure handling risk remains.",
+ "severity": "Medium",
+ "area": "Other",
+ "path": "apps/server/src/persistence/NodeSqliteClient.ts",
+ "line": 145,
+ "title": "In `executeUnprepared`, preparing outside an effect and not closing the statement can cause uncaught errors and leaks. Suggest wrapping with `Effect.acquireUseRelease`: prepare in acquire (error-safe) and `finalize` in release.",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5whxX9"
+ ]
+ },
+ {
+ "canonical_id": "C025",
+ "verdict": "invalid",
+ "rationale": "No longer duplicated; checkpoint ref helper now centralized.",
+ "severity": "Medium",
+ "area": "Checkpointing correctness",
+ "path": "apps/server/src/wsServer.ts",
+ "line": 128,
+ "title": "Duplicated checkpoint ref computation across two files",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wvwag"
+ ]
+ },
+ {
+ "canonical_id": "C026",
+ "verdict": "valid",
+ "rationale": "runPromise result still not caught; rejection can surface unhandled.",
+ "severity": "Medium",
+ "area": "WebSocket robustness",
+ "path": "apps/server/src/wsServer.ts",
+ "line": 545,
+ "title": "Consider adding `.catch(() => {})` after `Effect.runPromise(handleMessage(ws, raw))` to prevent unhandled rejections from crashing the server if `encodeResponse` or setup logic fails.",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wj4cE"
+ ]
+ },
+ {
+ "canonical_id": "C027",
+ "verdict": "valid",
+ "rationale": "Same unhandled rejection path remains in WS message handler.",
+ "severity": "Medium",
+ "area": "Runtime resilience and failure handling",
+ "path": "apps/server/src/wsServer.ts",
+ "line": 545,
+ "title": "WS message handler can cause unhandled promise rejection",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wyTaW",
+ "PRRT_kwDORLtfbc5wzli3"
+ ]
+ },
+ {
+ "canonical_id": "C028",
+ "verdict": "partially-valid",
+ "rationale": "Optimistic local+server dual update is intentional but can temporarily diverge.",
+ "severity": "Medium",
+ "area": "Other",
+ "path": "apps/web/src/components/BranchToolbar.tsx",
+ "line": 102,
+ "title": "Branch sync dispatches both server and stale local update",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5v-XCu"
+ ]
+ },
+ {
+ "canonical_id": "C029",
+ "verdict": "valid",
+ "rationale": "file URL decoding still double-decodes in one path.",
+ "severity": "Medium",
+ "area": "Edge-case parsing/platform behavior",
+ "path": "apps/web/src/markdown-links.ts",
+ "line": 105,
+ "title": "`parseFileUrlHref` already decodes the path (line 46), but `safeDecode` is called again here, corrupting filenames containing `%` sequences. Consider skipping the decode when `fileUrlTarget` is non-null.",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wnVsU"
+ ]
+ },
+ {
+ "canonical_id": "C030",
+ "verdict": "valid",
+ "rationale": "Scheme regex still misclassifies script.ts:10 as external scheme.",
+ "severity": "Medium",
+ "area": "Edge-case parsing/platform behavior",
+ "path": "apps/web/src/markdown-links.ts",
+ "line": 111,
+ "title": "`EXTERNAL_SCHEME_PATTERN` matches `script.ts:10` as a scheme because `.ts:` looks like `scheme:`. Consider requiring `://` after the colon, or checking that what follows the colon is not just digits. <details> <summary>\ud83d\ude80 Reply \"<strong>fix it for me</strong>\" or copy this <strong>AI Prompt</strong> for your agent:</summary>",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wnVsK"
+ ]
+ },
+ {
+ "canonical_id": "C031",
+ "verdict": "invalid",
+ "rationale": "Revert now uses explicit checkpointTurnCount first; positional fallback is non-primary.",
+ "severity": "Medium",
+ "area": "Event ordering and state consistency",
+ "path": "apps/web/src/session-logic.ts",
+ "line": 127,
+ "title": "Revert uses wrong turn count from positional inference",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5v9SCp"
+ ]
+ },
+ {
+ "canonical_id": "C032",
+ "verdict": "valid",
+ "rationale": "Dummy workflow marker comment is still present in docs.",
+ "severity": "Low",
+ "area": "Other",
+ "path": "AGENTS.md",
+ "line": 45,
+ "title": "Dummy workflow marker accidentally committed to docs",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wlYgj",
+ "PRRT_kwDORLtfbc5w1C4H"
+ ]
+ },
+ {
+ "canonical_id": "C033",
+ "verdict": "partially-valid",
+ "rationale": "Original claim overstated; some errors used, others appear unused.",
+ "severity": "Low",
+ "area": "Checkpointing correctness",
+ "path": "apps/server/src/checkpointing/Errors.ts",
+ "line": 51,
+ "title": "Three error classes defined but never instantiated anywhere",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wlYgo"
+ ]
+ },
+ {
+ "canonical_id": "C034",
+ "verdict": "valid",
+ "rationale": "CheckpointInvariantError remains redundantly included in service union.",
+ "severity": "Low",
+ "area": "Checkpointing correctness",
+ "path": "apps/server/src/checkpointing/Errors.ts",
+ "line": 79,
+ "title": "Redundant `CheckpointInvariantError` in `CheckpointServiceError` union type",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wj5fn"
+ ]
+ },
+ {
+ "canonical_id": "C035",
+ "verdict": "valid",
+ "rationale": "Same as C034.",
+ "severity": "Low",
+ "area": "Checkpointing correctness",
+ "path": "apps/server/src/checkpointing/Errors.ts",
+ "line": 79,
+ "title": "Redundant error type in CheckpointServiceError union definition",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wlYgs",
+ "PRRT_kwDORLtfbc5wxsO6",
+ "PRRT_kwDORLtfbc5w1C4B"
+ ]
+ },
+ {
+ "canonical_id": "C036",
+ "verdict": "invalid",
+ "rationale": "No longer duplicated; single production source via Refs.ts.",
+ "severity": "Low",
+ "area": "Checkpointing correctness",
+ "path": "apps/server/src/checkpointing/Layers/CheckpointStore.ts",
+ "line": 284,
+ "title": "Duplicate `checkpointRefForThreadTurn` function in two production files",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wiqFX"
+ ]
+ },
+ {
+ "canonical_id": "C037",
+ "verdict": "partially-valid",
+ "rationale": "Callback cleanup missing, but practical exposure is low in one-shot startup path.",
+ "severity": "Low",
+ "area": "Other",
+ "path": "apps/server/src/config.ts",
+ "line": 41,
+ "title": "`Effect.callback` should return a cleanup function to close the server(s) on fiber interruption. Without it, the `Net.Server` handles keep the process alive and leak the port if the effect is cancelled. <details> <summary>\ud83d\ude80 Reply \"<strong>fix it for me</strong>\" or copy this <strong>AI Prompt</strong> for your agent:</summary>",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wj4cO"
+ ]
+ },
+ {
+ "canonical_id": "C038",
+ "verdict": "valid",
+ "rationale": "Chunk-by-chunk UTF-8 decode can still corrupt split multibyte characters.",
+ "severity": "Low",
+ "area": "Edge-case parsing/platform behavior",
+ "path": "apps/server/src/git/Layers/CodexTextGeneration.ts",
+ "line": 136,
+ "title": "Multi-byte UTF-8 characters split across chunks will be corrupted when decoding each chunk separately. Consider accumulating all chunks first, then decoding once, or use `TextDecoder` with `stream: true`. <details> <summary>\ud83d\ude80 Reply \"<strong>fix it for me</strong>\" or copy this <strong>AI Prompt</strong> for your agent:</summary>",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5w1GPo"
+ ]
+ },
+ {
+ "canonical_id": "C039",
+ "verdict": "partially-valid",
+ "rationale": "Parser/encoder mismatch remains, but encoder path currently low-use.",
+ "severity": "Low",
+ "area": "Edge-case parsing/platform behavior",
+ "path": "apps/server/src/keybindings.ts",
+ "line": 352,
+ "title": "The `+` key can be parsed (via trailing `+` handling) but cannot be encoded because `shortcut.key.includes(\"+\")` returns true for the literal `+` key. Consider checking `shortcut.key === \"+\"` separately and encoding it as `\"space\"` style (e.g., a special token), or adjusting the condition to allow the single `+` character. <details> <summary>\ud83d\ude80 Reply \"<strong>fix it for me</strong>\" or copy this <strong>AI Prompt</strong> for your agent:</summary>",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wxvIB"
+ ]
+ },
+ {
+ "canonical_id": "C040",
+ "verdict": "valid",
+ "rationale": "upsertKeybindingRule read-modify-write remains race-prone.",
+ "severity": "Low",
+ "area": "Edge-case parsing/platform behavior",
+ "path": "apps/server/src/keybindings.ts",
+ "line": 488,
+ "title": "`upsertKeybindingRule` has a race condition: concurrent calls read the same file state, then the last write overwrites earlier changes. Consider wrapping the read-modify-write sequence with `Effect.Semaphore` to serialize access. <details> <summary>\ud83d\ude80 Reply \"<strong>fix it for me</strong>\" or copy this <strong>AI Prompt</strong> for your agent:</summary>",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wxvIA"
+ ]
+ },
+ {
+ "canonical_id": "C041",
+ "verdict": "valid",
+ "rationale": "Same Symbol interpolation hazard as C011 remains.",
+ "severity": "Low",
+ "area": "Other",
+ "path": "apps/server/src/orchestration/Errors.ts",
+ "line": 126,
+ "title": "Template literal interpolation throws `TypeError` if `cause` is a `Symbol`. Consider using `String(cause)` for safe coercion.",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wBFWs"
+ ]
+ },
+ {
+ "canonical_id": "C042",
+ "verdict": "partially-valid",
+ "rationale": "Duplication exists but one instance is variant logic, so impact is moderate.",
+ "severity": "Low",
+ "area": "Runtime resilience and failure handling",
+ "path": "apps/server/src/orchestration/Layers/CheckpointReactor.ts",
+ "line": 62,
+ "title": "Duplicated `resolveThreadWorkspaceCwd` across three files",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wzli2"
+ ]
+ },
+ {
+ "canonical_id": "C043",
+ "verdict": "valid",
+ "rationale": "Workspace CWD resolution duplication still present across modules.",
+ "severity": "Low",
+ "area": "Runtime resilience and failure handling",
+ "path": "apps/server/src/orchestration/Layers/CheckpointReactor.ts",
+ "line": 62,
+ "title": "Duplicated workspace CWD resolution logic across reactor modules",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wnWwM",
+ "PRRT_kwDORLtfbc5w1C3-",
+ "PRRT_kwDORLtfbc5w1HGZ"
+ ]
+ },
+ {
+ "canonical_id": "C044",
+ "verdict": "partially-valid",
+ "rationale": "Errors are swallowed to empty diff with warning; not fully silent but still lossy.",
+ "severity": "Low",
+ "area": "Runtime resilience and failure handling",
+ "path": "apps/server/src/orchestration/Layers/CheckpointReactor.ts",
+ "line": 274,
+ "title": "Checkpoint reactor swallows diff errors silently for `turn.completed`",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wkPaO"
+ ]
+ },
+ {
+ "canonical_id": "C045",
+ "verdict": "valid",
+ "rationale": "truncateDetail still overshoots limit.",
+ "severity": "Low",
+ "area": "Runtime resilience and failure handling",
+ "path": "apps/server/src/orchestration/Layers/ProviderRuntimeIngestion.ts",
+ "line": 29,
+ "title": "`truncateDetail` slices to `limit - 1` then appends `\"...\"` (3 chars), producing strings of length `limit + 2`. Consider slicing to `limit - 3` instead. <details> <summary>\ud83d\ude80 Reply \"<strong>fix it for me</strong>\" or copy this <strong>AI Prompt</strong> for your agent:</summary>",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wzp4R"
+ ]
+ },
+ {
+ "canonical_id": "C046",
+ "verdict": "valid",
+ "rationale": "latestMessageIdByTurnKey still unused/unpruned in per-turn clear path.",
+ "severity": "Low",
+ "area": "Runtime resilience and failure handling",
+ "path": "apps/server/src/orchestration/Layers/ProviderRuntimeIngestion.ts",
+ "line": 133,
+ "title": "`latestMessageIdByTurnKey` is written to but never read, and `clearAssistantMessageIdsForTurn` doesn't clear its entries\u2014only `clearTurnStateForSession` does. Consider removing this map entirely if unused, or clearing it alongside `turnMessageIdsByTurnKey` in `clearAssistantMessageIdsForTurn`. <details> <summary>\ud83d\ude80 Reply \"<strong>fix it for me</strong>\" or copy this <strong>AI Prompt</strong> for your agent:</summary>",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wxvIQ"
+ ]
+ },
+ {
+ "canonical_id": "C047",
+ "verdict": "valid",
+ "rationale": "Decode and SQL errors still collapsed into one persistence error kind.",
+ "severity": "Low",
+ "area": "Other",
+ "path": "apps/server/src/persistence/Layers/OrchestrationCommandReceipts.ts",
+ "line": 75,
+ "title": "`SqlSchema.findOneOption` can produce both SQL errors and decode errors, but `mapError` wraps all as `PersistenceSqlError`. Consider distinguishing `ParseError` from SQL errors and mapping decode failures to `PersistenceDecodeError` instead. <details> <summary>\ud83d\ude80 Reply \"<strong>fix it for me</strong>\" or copy this <strong>AI Prompt</strong> for your agent:</summary>",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wiaR-"
+ ]
+ },
+ {
+ "canonical_id": "C048",
+ "verdict": "partially-valid",
+ "rationale": "Unsafe run() result cast remains; runtime risk depends on raw-result consumers.",
+ "severity": "Low",
+ "area": "Other",
+ "path": "apps/server/src/persistence/NodeSqliteClient.ts",
+ "line": 99,
+ "title": "`statement.run()` returns a `RunResult` object, not an array. Casting it to `ReadonlyArray<any>` will cause runtime errors when consumers call array methods. Consider wrapping the result (e.g., `[result]`) or returning a properly typed object.",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wio-r"
+ ]
+ },
+ {
+ "canonical_id": "C049",
+ "verdict": "valid",
+ "rationale": "JSON.stringify(cause) may return undefined despite string expectations.",
+ "severity": "Low",
+ "area": "Other",
+ "path": "apps/server/src/provider/Layers/ProviderService.ts",
+ "line": 59,
+ "title": "`JSON.stringify(cause)` returns `undefined` for `undefined`, functions, or symbols, violating the `string` return type. Consider coercing the result to a string (e.g., `String(JSON.stringify(cause))`) or adding a fallback.",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wnVsI"
+ ]
+ },
+ {
+ "canonical_id": "C050",
+ "verdict": "valid",
+ "rationale": "ProviderSessionDirectory upsert remains read-merge-write without concurrency control.",
+ "severity": "Low",
+ "area": "Other",
+ "path": "apps/server/src/provider/Layers/ProviderSessionDirectory.ts",
+ "line": 94,
+ "title": "The read-modify-write pattern (`getBySessionId` \u2192 merge \u2192 `upsert`) is susceptible to lost updates under concurrent writes. Consider wrapping in a transaction or adding optimistic concurrency control (e.g., version field) if concurrent session updates are expected. <details> <summary>\ud83d\ude80 Reply \"<strong>fix it for me</strong>\" or copy this <strong>AI Prompt</strong> for your agent:</summary>",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wiLhY"
+ ]
+ },
+ {
+ "canonical_id": "C051",
+ "verdict": "partially-valid",
+ "rationale": "Null-clearing issue is real for providerThreadId; adapterKey part overstated.",
+ "severity": "Low",
+ "area": "Other",
+ "path": "apps/server/src/provider/Layers/ProviderSessionDirectory.ts",
+ "line": 119,
+ "title": "Using `??` for `providerThreadId` and `adapterKey` makes it impossible to clear these fields by passing `null`, since `null ?? existing` evaluates to `existing`. Consider using explicit `undefined` checks (like `resumeCursor` does) if clearing should be supported. <details> <summary>\ud83d\ude80 Reply \"<strong>fix it for me</strong>\" or copy this <strong>AI Prompt</strong> for your agent:</summary>",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wxvH9"
+ ]
+ },
+ {
+ "canonical_id": "C052",
+ "verdict": "valid",
+ "rationale": "Data callback may race before processHandle assignment.",
+ "severity": "Low",
+ "area": "Other",
+ "path": "apps/server/src/terminal/Layers/BunPTY.ts",
+ "line": 97,
+ "title": "Race condition: `processHandle` may be `null` when `data` callback fires, since it's assigned after `Bun.spawn` returns. Consider initializing `BunPtyProcess` first, then passing it to the callback to avoid losing initial output. <details> <summary>\ud83d\ude80 Reply \"<strong>fix it for me</strong>\" or copy this <strong>AI Prompt</strong> for your agent:</summary>",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5w1CxE"
+ ]
+ },
+ {
+ "canonical_id": "C053",
+ "verdict": "valid",
+ "rationale": "Still uses write+destroy rather than end() for rejection response.",
+ "severity": "Low",
+ "area": "WebSocket robustness",
+ "path": "apps/server/src/wsServer.ts",
+ "line": 83,
+ "title": "Consider using `socket.end(response)` instead of `socket.write(response)` + `socket.destroy()` to ensure the HTTP error response is fully flushed before closing the connection.",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5v-WPD"
+ ]
+ },
+ {
+ "canonical_id": "C054",
+ "verdict": "valid",
+ "rationale": "Array chunk UTF-8 decode remains vulnerable to split multibyte corruption.",
+ "severity": "Low",
+ "area": "WebSocket robustness",
+ "path": "apps/server/src/wsServer.ts",
+ "line": 104,
+ "title": "When array chunks contain a multi-byte UTF-8 character split across boundaries, decoding each chunk separately produces replacement characters. Consider using `Buffer.concat()` on all chunks before calling `.toString(\"utf8\")`. <details> <summary>\ud83d\ude80 Reply \"<strong>fix it for me</strong>\" or copy this <strong>AI Prompt</strong> for your agent:</summary>",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5whtrR"
+ ]
+ },
+ {
+ "canonical_id": "C055",
+ "verdict": "invalid",
+ "rationale": "No longer duplicated; helper is centralized.",
+ "severity": "Low",
+ "area": "Checkpointing correctness",
+ "path": "apps/server/src/wsServer.ts",
+ "line": 128,
+ "title": "Duplicate `checkpointRefForThreadTurn` function across files",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wkPaG"
+ ]
+ },
+ {
+ "canonical_id": "C056",
+ "verdict": "partially-valid",
+ "rationale": "Bug pattern exists, but current callsites mostly avoid triggering it.",
+ "severity": "Low",
+ "area": "Other",
+ "path": "apps/web/src/components/ui/sidebar.tsx",
+ "line": 114,
+ "title": "When `onOpenChange` is provided without `open`, the internal `_open` state never updates because `setOpenProp` takes precedence. Consider calling `_setOpen` when `openProp === undefined`, regardless of whether `setOpenProp` exists.",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wxvIq"
+ ]
+ },
+ {
+ "canonical_id": "C057",
+ "verdict": "valid",
+ "rationale": "Resizable object recreation still retriggers effect/storage reads.",
+ "severity": "Low",
+ "area": "Other",
+ "path": "apps/web/src/routes/_chat.$threadId.tsx",
+ "line": 105,
+ "title": "The `resizable` object is recreated on every render, causing `SidebarRail`'s `useEffect` to repeatedly read localStorage and update the DOM. Consider memoizing the object with `useMemo`. <details> <summary>\ud83d\ude80 Reply \"<strong>fix it for me</strong>\" or copy this <strong>AI Prompt</strong> for your agent:</summary>",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wyWz4"
+ ]
+ },
+ {
+ "canonical_id": "C058",
+ "verdict": "valid",
+ "rationale": "Number(null) -> 0 path still forces min width on initial load.",
+ "severity": "Low",
+ "area": "Other",
+ "path": "apps/web/src/routes/_chat.$threadId.tsx",
+ "line": 122,
+ "title": "When `localStorage.getItem()` returns `null`, `Number(null)` evaluates to `0`, which passes `Number.isFinite(0)`. This causes the sidebar to clamp to `minWidth` on first load, overriding the `DIFF_INLINE_DEFAULT_WIDTH` CSS clamp. Consider checking for `null` or empty string before parsing, e.g. guard with `storedWidth === null || storedWidth === ''`. <details> <summary>\ud83d\ude80 Reply \"<strong>fix it for me</strong>\" or copy this <strong>AI Prompt</strong> for your agent:</summary>",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5wnVsX"
+ ]
+ },
+ {
+ "canonical_id": "C059",
+ "verdict": "partially-valid",
+ "rationale": "Transport _tag override risk exists but current callsites are constrained.",
+ "severity": "Low",
+ "area": "WebSocket robustness",
+ "path": "apps/web/src/wsTransport.ts",
+ "line": 59,
+ "title": "Suggestion: don\u2019t spread `params` into `body`; it can override `_tag` and mishandle non-object values. Keep `_tag` separate and nest `params` under a single key (e.g., `data`), or validate `params` is a plain object.",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5whtrN"
+ ]
+ },
+ {
+ "canonical_id": "C060",
+ "verdict": "valid",
+ "rationale": "Schema still cannot express null clear for defaultModel patch.",
+ "severity": "Low",
+ "area": "Other",
+ "path": "packages/contracts/src/orchestration.ts",
+ "line": 253,
+ "title": "`defaultModel` should be `Schema.optional(Schema.NullOr(Schema.String))` to allow clearing the value. Currently there's no way to reset it to `null` since omitting means \"no change\" in patch semantics. <details> <summary>\ud83d\ude80 Reply \"<strong>fix it for me</strong>\" or copy this <strong>AI Prompt</strong> for your agent:</summary>",
+ "thread_ids": [
+ "PRRT_kwDORLtfbc5whxJC"
+ ]
+ }
+ ]
+}
diff --git a/.plans/16b-pr89-architecture-validity-audit.md b/.plans/16b-pr89-architecture-validity-audit.md
new file mode 100644
index 000000000..946bc2270
--- /dev/null
+++ b/.plans/16b-pr89-architecture-validity-audit.md
@@ -0,0 +1,125 @@
+# PR #89 Post-Rewrite Validity Audit
+
+Architecture baseline used for this audit:
+- Provider runtime activity is projected into orchestration domain events.
+- Web client consumes `orchestration.domainEvent`; no separate `providers.event` channel is expected.
+
+## Summary
+
+- Canonical items reviewed: 58
+- Valid: 34
+- Partially valid: 19
+- Invalid (stale/no longer applicable): 5
+
+## Invalid Items To Close
+
+- C021 [Medium] Shared mutable default metadata object causes stale eventId (apps/server/src/orchestration/decider.ts:27)
+ - Reason: Stale-eventId claim no longer applies; eventId is regenerated per event.
+ - Threads: PRRT_kwDORLtfbc5wkPaA
+- C025 [Medium] Duplicated checkpoint ref computation across two files (apps/server/src/wsServer.ts:128)
+ - Reason: No longer duplicated; checkpoint ref helper now centralized.
+ - Threads: PRRT_kwDORLtfbc5wvwag
+- C031 [Medium] Revert uses wrong turn count from positional inference (apps/web/src/session-logic.ts:127)
+ - Reason: Revert now uses explicit checkpointTurnCount first; positional fallback is non-primary.
+ - Threads: PRRT_kwDORLtfbc5v9SCp
+- C036 [Low] Duplicate `checkpointRefForThreadTurn` function in two production files (apps/server/src/checkpointing/Layers/CheckpointStore.ts:284)
+ - Reason: No longer duplicated; single production source via Refs.ts.
+ - Threads: PRRT_kwDORLtfbc5wiqFX
+- C055 [Low] Duplicate `checkpointRefForThreadTurn` function across files (apps/server/src/wsServer.ts:128)
+ - Reason: No longer duplicated; helper is centralized.
+ - Threads: PRRT_kwDORLtfbc5wkPaG
+
+## Partially Valid Items (de-scope or lower priority)
+
+- C006 [Medium] Build script runs TypeScript file with bare node (apps/server/package.json:14)
+ - Note: Node 24 makes this work, but script still relies on runtime-specific TS execution.
+- C007 [Medium] Build script uses `node` to run `.ts` file directly (apps/server/package.json:14)
+ - Note: Same as C006 (duplicate concern).
+- C008 [Medium] Inconsistent input normalization across CheckpointStore methods (apps/server/src/checkpointing/Layers/CheckpointStore.ts:94)
+ - Note: Normalization inconsistency remains, but upstream callsites now mostly canonicalize inputs.
+- C014 [Medium] Engine error handler catches all errors including non-invariant ones (apps/server/src/orchestration/Layers/OrchestrationEngine.ts:144)
+ - Note: Broad catch may be intentional for liveness; correctness concern is contextual.
+- C015 [Medium] The gap-filling fallback logic can retain messages from turns that are about to be deleted, causing foreign key violations. Consider removing the fallback logic entirely, or filtering `fallbackUserMessages` and `fallbackAssistantMessages` to only include messages whose `turnId` is in `retainedTurnIds`. <details> <summary>🚀 Reply "<strong>fix it for me</strong>" or copy this <strong>AI Prompt</strong> for your agent:</summary> (apps/server/src/orchestration/Layers/ProjectionPipeline.ts:99)
+ - Note: Message fallback retention issue is real, but prior FK-violation claim is overstated.
+- C017 [Medium] `REQUIRED_SNAPSHOT_PROJECTORS` includes `pending-approvals` and `thread-turns`, but `getSnapshot` doesn't query their data. If these projectors lag behind, the returned `snapshotSequence` will be lower than what the included data actually reflects, causing clients to replay already-applied events. Consider filtering `REQUIRED_SNAPSHOT_PROJECTORS` to only include projectors whose data is actually fetched in the snapshot. <details> <summary>🚀 Reply "<strong>fix it for me</strong>" or copy this <strong>AI Prompt</strong> for your agent:</summary> (apps/server/src/orchestration/Layers/ProjectionSnapshotQuery.ts:71)
+ - Note: Snapshot sequence can under-report due to extra projectors, but replay impact is lower now.
+- C019 [Medium] ProviderRuntimeIngestion processes events for wrong thread on race (apps/server/src/orchestration/Layers/ProviderRuntimeIngestion.ts:178)
+ - Note: SessionId-only routing can misassociate events under races/rebinds.
+- C020 [Medium] On `message.completed`, the message ID is added to the set and `thread.message.assistant.complete` is dispatched. On `turn.completed`, the same set is iterated and `thread.message.assistant.complete` is dispatched again for each ID—including already-completed ones. Consider removing message IDs from the set after dispatching on `message.completed`, or filtering out already-completed IDs before the `turn.completed` loop. (apps/server/src/orchestration/Layers/ProviderRuntimeIngestion.ts:266)
+ - Note: Duplicate complete dispatch exists; downstream impact often idempotent.
+- C024 [Medium] In `executeUnprepared`, preparing outside an effect and not closing the statement can cause uncaught errors and leaks. Suggest wrapping with `Effect.acquireUseRelease`: prepare in acquire (error-safe) and `finalize` in release. (apps/server/src/persistence/NodeSqliteClient.ts:145)
+ - Note: prepare() still outside managed effect boundary; failure handling risk remains.
+- C028 [Medium] Branch sync dispatches both server and stale local update (apps/web/src/components/BranchToolbar.tsx:102)
+ - Note: Optimistic local+server dual update is intentional but can temporarily diverge.
+- C033 [Low] Three error classes defined but never instantiated anywhere (apps/server/src/checkpointing/Errors.ts:51)
+ - Note: Original claim overstated; some errors used, others appear unused.
+- C037 [Low] `Effect.callback` should return a cleanup function to close the server(s) on fiber interruption. Without it, the `Net.Server` handles keep the process alive and leak the port if the effect is cancelled. <details> <summary>🚀 Reply "<strong>fix it for me</strong>" or copy this <strong>AI Prompt</strong> for your agent:</summary> (apps/server/src/config.ts:41)
+ - Note: Callback cleanup missing, but practical exposure is low in one-shot startup path.
+- C039 [Low] The `+` key can be parsed (via trailing `+` handling) but cannot be encoded because `shortcut.key.includes("+")` returns true for the literal `+` key. Consider checking `shortcut.key === "+"` separately and encoding it as `"space"` style (e.g., a special token), or adjusting the condition to allow the single `+` character. <details> <summary>🚀 Reply "<strong>fix it for me</strong>" or copy this <strong>AI Prompt</strong> for your agent:</summary> (apps/server/src/keybindings.ts:352)
+ - Note: Parser/encoder mismatch remains, but encoder path currently low-use.
+- C042 [Low] Duplicated `resolveThreadWorkspaceCwd` across three files (apps/server/src/orchestration/Layers/CheckpointReactor.ts:62)
+ - Note: Duplication exists but one instance is variant logic, so impact is moderate.
+- C044 [Low] Checkpoint reactor swallows diff errors silently for `turn.completed` (apps/server/src/orchestration/Layers/CheckpointReactor.ts:274)
+ - Note: Errors are swallowed to empty diff with warning; not fully silent but still lossy.
+- C048 [Low] `statement.run()` returns a `RunResult` object, not an array. Casting it to `ReadonlyArray<any>` will cause runtime errors when consumers call array methods. Consider wrapping the result (e.g., `[result]`) or returning a properly typed object. (apps/server/src/persistence/NodeSqliteClient.ts:99)
+ - Note: Unsafe run() result cast remains; runtime risk depends on raw-result consumers.
+- C051 [Low] Using `??` for `providerThreadId` and `adapterKey` makes it impossible to clear these fields by passing `null`, since `null ?? existing` evaluates to `existing`. Consider using explicit `undefined` checks (like `resumeCursor` does) if clearing should be supported. <details> <summary>🚀 Reply "<strong>fix it for me</strong>" or copy this <strong>AI Prompt</strong> for your agent:</summary> (apps/server/src/provider/Layers/ProviderSessionDirectory.ts:119)
+ - Note: Null-clearing issue is real for providerThreadId; adapterKey part overstated.
+- C056 [Low] When `onOpenChange` is provided without `open`, the internal `_open` state never updates because `setOpenProp` takes precedence. Consider calling `_setOpen` when `openProp === undefined`, regardless of whether `setOpenProp` exists. (apps/web/src/components/ui/sidebar.tsx:114)
+ - Note: Bug pattern exists, but current callsites mostly avoid triggering it.
+- C059 [Low] Suggestion: don’t spread `params` into `body`; it can override `_tag` and mishandle non-object values. Keep `_tag` separate and nest `params` under a single key (e.g., `data`), or validate `params` is a plain object. (apps/web/src/wsTransport.ts:59)
+ - Note: Transport _tag override risk exists but current callsites are constrained.
+
+## Valid Items (actionable unless marked completed)
+
+- C001 [High] Non-atomic event appending can corrupt state on retry. If an error occurs mid-loop (lines 96-102) after some events are persisted but before the receipt is written, the command appears to fail. A retry generates new UUIDs via `crypto.randomUUID()` in the decider, appending duplicate events. Consider wrapping the loop in a transaction or using deterministic event IDs derived from `commandId`. <details> <summary>🚀 Reply "<strong>fix it for me</strong>" or copy this <strong>AI Prompt</strong> for your agent:</summary> (apps/server/src/orchestration/Layers/OrchestrationEngine.ts:96)
+- C002 [High] A dispatch error in `processEvent` will terminate the `Effect.forever` loop, permanently halting event ingestion. Consider adding error recovery (e.g., `Effect.catchAll` with logging) around `processEvent` so failures don't kill the fiber. (apps/server/src/orchestration/Layers/ProviderRuntimeIngestion.ts:333)
+ - Status: Completed in Phase 1.
+- C003 [High] Consider attaching a no-op error listener before `socket.write` (e.g., `socket.on('error', () => {})`) to prevent an unhandled `EPIPE`/`ECONNRESET` from crashing the process if the client disconnects mid-handshake. (apps/server/src/wsServer.ts:75)
+- C009 [Medium] Git's braced rename syntax (e.g., `src/{old => new}/file.ts`) isn't handled correctly. The current slice after ` => ` produces invalid paths like `new}/file.ts`. Consider expanding the braces to construct the full destination path. <details> <summary>🚀 Reply "<strong>fix it for me</strong>" or copy this <strong>AI Prompt</strong> for your agent:</summary> (apps/server/src/git/Layers/GitCore.ts:41)
+- C010 [Medium] `loadCustomKeybindingsConfig` fails when the config file doesn't exist, which is expected for new users. Consider catching `ENOENT` and returning an empty array instead. <details> <summary>🚀 Reply "<strong>fix it for me</strong>" or copy this <strong>AI Prompt</strong> for your agent:</summary> (apps/server/src/keybindings.ts:418)
+- C011 [Medium] Template literal `${cause}` throws if `cause` is a `Symbol`. Consider using `String(cause)` instead. (apps/server/src/orchestration/Errors.ts:117)
+- C012 [Medium] Forked revert dispatch risks read model inconsistency (apps/server/src/orchestration/Layers/CheckpointReactor.ts:542)
+ - Status: Completed in Phase 1.
+- C013 [Medium] If `projectionPipeline.projectEvent` fails after `eventStore.append` succeeds, the event is persisted but `readModel` isn't updated, causing desync. Consider updating the in-memory `readModel` immediately after append (before the external projection), so local state stays consistent regardless of downstream failures. (apps/server/src/orchestration/Layers/OrchestrationEngine.ts:99)
+- C016 [Medium] The in-memory `pendingTurnStartByThreadId` map isn't restored during bootstrap. If the service restarts after processing `thread.turn-start-requested` but before `thread.session-set`, the `userMessageId` and `startedAt` will be lost since bootstrap resumes *after* the committed sequence. Consider persisting this pending state or processing these two events atomically. <details> <summary>🚀 Reply "<strong>fix it for me</strong>" or copy this <strong>AI Prompt</strong> for your agent:</summary> (apps/server/src/orchestration/Layers/ProjectionPipeline.ts:490)
+- C018 [Medium] Unbounded memory growth in turn start deduplication set (apps/server/src/orchestration/Layers/ProviderCommandReactor.ts:84)
+- C022 [Medium] Fish shell outputs `$PATH` as space-separated, not colon-separated. Consider checking if the shell is fish and using `string join : $PATH` instead, or validating the result contains colons before assigning. (apps/server/src/os-jank.ts:10)
+- C023 [Medium] Using `-il` flags causes the shell to source profile scripts that may print banners or other text, polluting the captured `PATH`. Consider using `-lc` (login only, non-interactive) to reduce unwanted output. (apps/server/src/os-jank.ts:10)
+- C026 [Medium] Consider adding `.catch(() => {})` after `Effect.runPromise(handleMessage(ws, raw))` to prevent unhandled rejections from crashing the server if `encodeResponse` or setup logic fails. (apps/server/src/wsServer.ts:545)
+ - Status: Completed in Phase 1.
+- C027 [Medium] WS message handler can cause unhandled promise rejection (apps/server/src/wsServer.ts:545)
+ - Status: Completed in Phase 1.
+- C029 [Medium] `parseFileUrlHref` already decodes the path (line 46), but `safeDecode` is called again here, corrupting filenames containing `%` sequences. Consider skipping the decode when `fileUrlTarget` is non-null. (apps/web/src/markdown-links.ts:105)
+- C030 [Medium] `EXTERNAL_SCHEME_PATTERN` matches `script.ts:10` as a scheme because `.ts:` looks like `scheme:`. Consider requiring `://` after the colon, or checking that what follows the colon is not just digits. <details> <summary>🚀 Reply "<strong>fix it for me</strong>" or copy this <strong>AI Prompt</strong> for your agent:</summary> (apps/web/src/markdown-links.ts:111)
+- C032 [Low] Dummy workflow marker accidentally committed to docs (AGENTS.md:45)
+- C034 [Low] Redundant `CheckpointInvariantError` in `CheckpointServiceError` union type (apps/server/src/checkpointing/Errors.ts:79)
+- C035 [Low] Redundant error type in CheckpointServiceError union definition (apps/server/src/checkpointing/Errors.ts:79)
+- C038 [Low] Multi-byte UTF-8 characters split across chunks will be corrupted when decoding each chunk separately. Consider accumulating all chunks first, then decoding once, or use `TextDecoder` with `stream: true`. <details> <summary>🚀 Reply "<strong>fix it for me</strong>" or copy this <strong>AI Prompt</strong> for your agent:</summary> (apps/server/src/git/Layers/CodexTextGeneration.ts:136)
+- C040 [Low] `upsertKeybindingRule` has a race condition: concurrent calls read the same file state, then the last write overwrites earlier changes. Consider wrapping the read-modify-write sequence with `Effect.Semaphore` to serialize access. <details> <summary>🚀 Reply "<strong>fix it for me</strong>" or copy this <strong>AI Prompt</strong> for your agent:</summary> (apps/server/src/keybindings.ts:488)
+- C041 [Low] Template literal interpolation throws `TypeError` if `cause` is a `Symbol`. Consider using `String(cause)` for safe coercion. (apps/server/src/orchestration/Errors.ts:126)
+- C043 [Low] Duplicated workspace CWD resolution logic across reactor modules (apps/server/src/orchestration/Layers/CheckpointReactor.ts:62)
+- C045 [Low] `truncateDetail` slices to `limit - 1` then appends `"..."` (3 chars), producing strings of length `limit + 2`. Consider slicing to `limit - 3` instead. <details> <summary>🚀 Reply "<strong>fix it for me</strong>" or copy this <strong>AI Prompt</strong> for your agent:</summary> (apps/server/src/orchestration/Layers/ProviderRuntimeIngestion.ts:29)
+- C046 [Low] `latestMessageIdByTurnKey` is written to but never read, and `clearAssistantMessageIdsForTurn` doesn't clear its entries—only `clearTurnStateForSession` does. Consider removing this map entirely if unused, or clearing it alongside `turnMessageIdsByTurnKey` in `clearAssistantMessageIdsForTurn`. <details> <summary>🚀 Reply "<strong>fix it for me</strong>" or copy this <strong>AI Prompt</strong> for your agent:</summary> (apps/server/src/orchestration/Layers/ProviderRuntimeIngestion.ts:133)
+- C047 [Low] `SqlSchema.findOneOption` can produce both SQL errors and decode errors, but `mapError` wraps all as `PersistenceSqlError`. Consider distinguishing `ParseError` from SQL errors and mapping decode failures to `PersistenceDecodeError` instead. <details> <summary>🚀 Reply "<strong>fix it for me</strong>" or copy this <strong>AI Prompt</strong> for your agent:</summary> (apps/server/src/persistence/Layers/OrchestrationCommandReceipts.ts:75)
+- C049 [Low] `JSON.stringify(cause)` returns `undefined` for `undefined`, functions, or symbols, violating the `string` return type. Consider coercing the result to a string (e.g., `String(JSON.stringify(cause))`) or adding a fallback. (apps/server/src/provider/Layers/ProviderService.ts:59)
+- C050 [Low] The read-modify-write pattern (`getBySessionId` → merge → `upsert`) is susceptible to lost updates under concurrent writes. Consider wrapping in a transaction or adding optimistic concurrency control (e.g., version field) if concurrent session updates are expected. <details> <summary>🚀 Reply "<strong>fix it for me</strong>" or copy this <strong>AI Prompt</strong> for your agent:</summary> (apps/server/src/provider/Layers/ProviderSessionDirectory.ts:94)
+- C052 [Low] Race condition: `processHandle` may be `null` when `data` callback fires, since it's assigned after `Bun.spawn` returns. Consider initializing `BunPtyProcess` first, then passing it to the callback to avoid losing initial output. <details> <summary>🚀 Reply "<strong>fix it for me</strong>" or copy this <strong>AI Prompt</strong> for your agent:</summary> (apps/server/src/terminal/Layers/BunPTY.ts:97)
+- C053 [Low] Consider using `socket.end(response)` instead of `socket.write(response)` + `socket.destroy()` to ensure the HTTP error response is fully flushed before closing the connection. (apps/server/src/wsServer.ts:83)
+- C054 [Low] When array chunks contain a multi-byte UTF-8 character split across boundaries, decoding each chunk separately produces replacement characters. Consider using `Buffer.concat()` on all chunks before calling `.toString("utf8")`. <details> <summary>🚀 Reply "<strong>fix it for me</strong>" or copy this <strong>AI Prompt</strong> for your agent:</summary> (apps/server/src/wsServer.ts:104)
+- C057 [Low] The `resizable` object is recreated on every render, causing `SidebarRail`'s `useEffect` to repeatedly read localStorage and update the DOM. Consider memoizing the object with `useMemo`. <details> <summary>🚀 Reply "<strong>fix it for me</strong>" or copy this <strong>AI Prompt</strong> for your agent:</summary> (apps/web/src/routes/_chat.$threadId.tsx:105)
+- C058 [Low] When `localStorage.getItem()` returns `null`, `Number(null)` evaluates to `0`, which passes `Number.isFinite(0)`. This causes the sidebar to clamp to `minWidth` on first load, overriding the `DIFF_INLINE_DEFAULT_WIDTH` CSS clamp. Consider checking for `null` or empty string before parsing, e.g. guard with `storedWidth === null || storedWidth === ''`. <details> <summary>🚀 Reply "<strong>fix it for me</strong>" or copy this <strong>AI Prompt</strong> for your agent:</summary> (apps/web/src/routes/_chat.$threadId.tsx:122)
+- C060 [Low] `defaultModel` should be `Schema.optional(Schema.NullOr(Schema.String))` to allow clearing the value. Currently there's no way to reset it to `null` since omitting means "no change" in patch semantics. <details> <summary>🚀 Reply "<strong>fix it for me</strong>" or copy this <strong>AI Prompt</strong> for your agent:</summary> (packages/contracts/src/orchestration.ts:253)
+
+## By Area
+
+- Build/runtime portability: valid=0, partially-valid=2, invalid=0
+- Checkpointing correctness: valid=2, partially-valid=3, invalid=3
+- Edge-case parsing/platform behavior: valid=8, partially-valid=1, invalid=0
+- Event ordering and state consistency: valid=3, partially-valid=2, invalid=1
+- Memory/resource growth: valid=1, partially-valid=0, invalid=0
+- Other: valid=10, partially-valid=6, invalid=1
+- Runtime resilience and failure handling: valid=6, partially-valid=4, invalid=0
+- WebSocket robustness: valid=4, partially-valid=1, invalid=0
+
+Machine-readable mapping: `.plans/16b-pr89-architecture-validity-audit.json`.
diff --git a/apps/server/src/orchestration/Layers/CheckpointReactor.test.ts b/apps/server/src/orchestration/Layers/CheckpointReactor.test.ts
index 99b531db0..2f5fa8671 100644
--- a/apps/server/src/orchestration/Layers/CheckpointReactor.test.ts
+++ b/apps/server/src/orchestration/Layers/CheckpointReactor.test.ts
@@ -42,10 +42,12 @@ const asSessionId = (value: string): ProviderSessionId => ProviderSessionId.make
const asProviderThreadId = (value: string): ProviderThreadId => ProviderThreadId.makeUnsafe(value);
const asTurnId = (value: string): TurnId => TurnId.makeUnsafe(value);
-function createProviderServiceHarness(cwd: string, hasSession = true) {
+function createProviderServiceHarness(cwd: string, hasSession = true, sessionCwd = cwd) {
const now = new Date().toISOString();
const runtimeEventPubSub = Effect.runSync(PubSub.unbounded<ProviderRuntimeEvent>());
- const rollbackConversation = vi.fn(() => Effect.void);
+ const rollbackConversation = vi.fn(
+ (_input: { readonly sessionId: ProviderSessionId; readonly numTurns: number }) => Effect.void,
+ );
const unsupported = <A>() =>
Effect.die(new Error("Unsupported provider call in test")) as Effect.Effect<A, never>;
@@ -57,7 +59,7 @@ function createProviderServiceHarness(cwd: string, hasSession = true) {
provider: "codex",
status: "ready",
threadId: asProviderThreadId("provider-thread-1"),
- cwd,
+ cwd: sessionCwd,
createdAt: now,
updatedAt: now,
},
@@ -214,10 +216,15 @@ describe("CheckpointReactor", () => {
readonly seedFilesystemCheckpoints?: boolean;
readonly projectWorkspaceRoot?: string;
readonly threadWorktreePath?: string | null;
+ readonly providerSessionCwd?: string;
}) {
const cwd = createGitRepository();
tempDirs.push(cwd);
- const provider = createProviderServiceHarness(cwd, options?.hasSession ?? true);
+ const provider = createProviderServiceHarness(
+ cwd,
+ options?.hasSession ?? true,
+ options?.providerSessionCwd ?? cwd,
+ );
const orchestrationLayer = OrchestrationEngineLive.pipe(
Layer.provide(OrchestrationProjectionPipelineLive),
Layer.provide(OrchestrationEventStoreLive),
@@ -496,6 +503,65 @@ describe("CheckpointReactor", () => {
);
});
+ it("continues processing runtime events after a single checkpoint runtime failure", async () => {
+ const nonRepositorySessionCwd = fs.mkdtempSync(
+ path.join(os.tmpdir(), "t3-checkpoint-runtime-non-repo-"),
+ );
+ tempDirs.push(nonRepositorySessionCwd);
+
+ const harness = await createHarness({
+ seedFilesystemCheckpoints: false,
+ providerSessionCwd: nonRepositorySessionCwd,
+ });
+ const createdAt = new Date().toISOString();
+
+ await Effect.runPromise(
+ harness.engine.dispatch({
+ type: "thread.session.set",
+ commandId: CommandId.makeUnsafe("cmd-session-set-non-repo-runtime"),
+ threadId: ThreadId.makeUnsafe("thread-1"),
+ session: {
+ threadId: ThreadId.makeUnsafe("thread-1"),
+ status: "ready",
+ providerName: "codex",
+ providerSessionId: asSessionId("sess-1"),
+ providerThreadId: ProviderThreadId.makeUnsafe("provider-thread-1"),
+ activeTurnId: null,
+ lastError: null,
+ updatedAt: createdAt,
+ },
+ createdAt,
+ }),
+ );
+
+ harness.provider.emit({
+ type: "checkpoint.captured",
+ eventId: EventId.makeUnsafe("evt-runtime-capture-failure"),
+ provider: "codex",
+ sessionId: asSessionId("sess-1"),
+ createdAt: new Date().toISOString(),
+ threadId: ProviderThreadId.makeUnsafe("provider-thread-1"),
+ turnId: asTurnId("turn-runtime-failure"),
+ turnCount: 1,
+ status: "completed",
+ });
+
+ harness.provider.emit({
+ type: "turn.started",
+ eventId: EventId.makeUnsafe("evt-turn-started-after-runtime-failure"),
+ provider: "codex",
+ sessionId: asSessionId("sess-1"),
+ createdAt: new Date().toISOString(),
+ threadId: ProviderThreadId.makeUnsafe("provider-thread-1"),
+ turnId: asTurnId("turn-after-runtime-failure"),
+ });
+
+ await waitForGitRefExists(harness.cwd, checkpointRefForThreadTurn(ThreadId.makeUnsafe("thread-1"), 0));
+ expect(
+ gitRefExists(harness.cwd, checkpointRefForThreadTurn(ThreadId.makeUnsafe("thread-1"), 0)),
+ ).toBe(true);
+ });
+
it("executes provider revert and emits thread.reverted for checkpoint revert requests", async () => {
const harness = await createHarness();
const createdAt = new Date().toISOString();
@@ -575,6 +641,101 @@ describe("CheckpointReactor", () => {
);
});
+ it("processes consecutive revert requests with deterministic rollback sequencing", async () => {
+ const harness = await createHarness();
+ const createdAt = new Date().toISOString();
+
+ await Effect.runPromise(
+ harness.engine.dispatch({
+ type: "thread.session.set",
+ commandId: CommandId.makeUnsafe("cmd-session-set-inline-revert"),
+ threadId: ThreadId.makeUnsafe("thread-1"),
+ session: {
+ threadId: ThreadId.makeUnsafe("thread-1"),
+ status: "ready",
+ providerName: "codex",
+ providerSessionId: asSessionId("sess-1"),
+ providerThreadId: ProviderThreadId.makeUnsafe("provider-thread-1"),
+ activeTurnId: null,
+ lastError: null,
+ updatedAt: createdAt,
+ },
+ createdAt,
+ }),
+ );
+
+ await Effect.runPromise(
+ harness.engine.dispatch({
+ type: "thread.turn.diff.complete",
+ commandId: CommandId.makeUnsafe("cmd-inline-revert-diff-1"),
+ threadId: ThreadId.makeUnsafe("thread-1"),
+ turnId: asTurnId("turn-1"),
+ completedAt: createdAt,
+ checkpointRef: checkpointRefForThreadTurn(ThreadId.makeUnsafe("thread-1"), 1),
+ status: "ready",
+ files: [],
+ checkpointTurnCount: 1,
+ createdAt,
+ }),
+ );
+ await Effect.runPromise(
+ harness.engine.dispatch({
+ type: "thread.turn.diff.complete",
+ commandId: CommandId.makeUnsafe("cmd-inline-revert-diff-2"),
+ threadId: ThreadId.makeUnsafe("thread-1"),
+ turnId: asTurnId("turn-2"),
+ completedAt: createdAt,
+ checkpointRef: checkpointRefForThreadTurn(ThreadId.makeUnsafe("thread-1"), 2),
+ status: "ready",
+ files: [],
+ checkpointTurnCount: 2,
+ createdAt,
+ }),
+ );
+
+ await Effect.runPromise(
+ harness.engine.dispatch({
+ type: "thread.checkpoint.revert",
+ commandId: CommandId.makeUnsafe("cmd-sequenced-revert-request-1"),
+ threadId: ThreadId.makeUnsafe("thread-1"),
+ turnCount: 1,
+ createdAt,
+ }),
+ );
+ await Effect.runPromise(
+ harness.engine.dispatch({
+ type: "thread.checkpoint.revert",
+ commandId: CommandId.makeUnsafe("cmd-sequenced-revert-request-0"),
+ threadId: ThreadId.makeUnsafe("thread-1"),
+ turnCount: 0,
+ createdAt,
+ }),
+ );
+
+ const deadline = Date.now() + 2000;
+ const waitForRollbackCalls = async (): Promise<void> => {
+ if (harness.provider.rollbackConversation.mock.calls.length >= 2) {
+ return;
+ }
+ if (Date.now() >= deadline) {
+ throw new Error("Timed out waiting for rollbackConversation calls.");
+ }
+ await new Promise((resolve) => setTimeout(resolve, 10));
+ return waitForRollbackCalls();
+ };
+ await waitForRollbackCalls();
+
+ expect(harness.provider.rollbackConversation).toHaveBeenCalledTimes(2);
+ expect(harness.provider.rollbackConversation.mock.calls[0]?.[0]).toEqual({
+ sessionId: asSessionId("sess-1"),
+ numTurns: 1,
+ });
+ expect(harness.provider.rollbackConversation.mock.calls[1]?.[0]).toEqual({
+ sessionId: asSessionId("sess-1"),
+ numTurns: 1,
+ });
+ });
+
it("appends an error activity when revert is requested without an active session", async () => {
const harness = await createHarness({ hasSession: false });
const createdAt = new Date().toISOString();
diff --git a/apps/server/src/orchestration/Layers/CheckpointReactor.ts b/apps/server/src/orchestration/Layers/CheckpointReactor.ts
index 234bb6f95..a0fa7dacc 100644
--- a/apps/server/src/orchestration/Layers/CheckpointReactor.ts
+++ b/apps/server/src/orchestration/Layers/CheckpointReactor.ts
@@ -9,7 +9,7 @@ import {
type OrchestrationEvent,
type ProviderRuntimeEvent,
} from "@t3tools/contracts";
-import { Effect, Layer, Option, Queue, Stream } from "effect";
+import { Cause, Effect, Layer, Option, Queue, Stream } from "effect";
import { parseTurnDiffFilesFromUnifiedDiff } from "../../checkpointing/Diffs.ts";
import { checkpointRefForThreadTurn } from "../../checkpointing/Refs.ts";
@@ -537,7 +537,6 @@ const make = Effect.gen(function* () {
createdAt: now,
}),
),
- Effect.forkScoped,
Effect.asVoid,
);
});
@@ -672,11 +671,27 @@ const make = Effect.gen(function* () {
): Effect.Effect<void, CheckpointStoreError | OrchestrationDispatchError, never> =>
input.source === "domain" ? processDomainEvent(input.event) : processRuntimeEvent(input.event);
+ const processInputSafely = (input: ReactorInput) =>
+ processInput(input).pipe(
+ Effect.catchCause((cause) => {
+ if (Cause.hasInterruptsOnly(cause)) {
+ return Effect.failCause(cause);
+ }
+ return Effect.logWarning("checkpoint reactor failed to process input", {
+ source: input.source,
+ eventType: input.event.type,
+ cause: Cause.pretty(cause),
+ });
+ }),
+ );
+
const start: CheckpointReactorShape["start"] = Effect.gen(function* () {
const queue = yield* Queue.unbounded<ReactorInput>();
yield* Effect.addFinalizer(() => Queue.shutdown(queue).pipe(Effect.asVoid));
- yield* Effect.forkScoped(Effect.forever(Queue.take(queue).pipe(Effect.flatMap(processInput))));
+ yield* Effect.forkScoped(
+ Effect.forever(Queue.take(queue).pipe(Effect.flatMap(processInputSafely))),
+ );
yield* Effect.forkScoped(
Stream.runForEach(orchestrationEngine.streamDomainEvents, (event) => {
diff --git a/apps/server/src/orchestration/Layers/ProviderRuntimeIngestion.test.ts b/apps/server/src/orchestration/Layers/ProviderRuntimeIngestion.test.ts
index d39fe290b..df97787a9 100644
--- a/apps/server/src/orchestration/Layers/ProviderRuntimeIngestion.test.ts
+++ b/apps/server/src/orchestration/Layers/ProviderRuntimeIngestion.test.ts
@@ -327,4 +327,40 @@ describe("ProviderRuntimeIngestion", () => {
expect(thread.session?.status).toBe("ready");
expect(thread.activities.some((activity) => activity.kind === "tool.started")).toBe(true);
});
+
+ it("continues processing runtime events after a single event handler failure", async () => {
+ const harness = await createHarness();
+ const now = new Date().toISOString();
+
+ harness.emit({
+ type: "message.delta",
+ eventId: asEventId("evt-invalid-delta"),
+ provider: "codex",
+ sessionId: asSessionId("sess-1"),
+ createdAt: now,
+ turnId: asProviderTurnId("turn-invalid"),
+ itemId: asItemId("item-invalid"),
+ delta: undefined,
+ } as unknown as ProviderRuntimeEvent);
+
+ harness.emit({
+ type: "runtime.error",
+ eventId: asEventId("evt-runtime-error-after-failure"),
+ provider: "codex",
+ sessionId: asSessionId("sess-1"),
+ createdAt: new Date().toISOString(),
+ turnId: asProviderTurnId("turn-after-failure"),
+ message: "runtime still processed",
+ });
+
+ const thread = await waitForThread(
+ harness.engine,
+ (entry) =>
+ entry.session?.status === "error" &&
+ entry.session?.activeTurnId === "turn-after-failure" &&
+ entry.session?.lastError === "runtime still processed",
+ );
+ expect(thread.session?.status).toBe("error");
+ expect(thread.session?.lastError).toBe("runtime still processed");
+ });
});
diff --git a/apps/server/src/orchestration/Layers/ProviderRuntimeIngestion.ts b/apps/server/src/orchestration/Layers/ProviderRuntimeIngestion.ts
index a3ab7e7c7..7ac968384 100644
--- a/apps/server/src/orchestration/Layers/ProviderRuntimeIngestion.ts
+++ b/apps/server/src/orchestration/Layers/ProviderRuntimeIngestion.ts
@@ -7,7 +7,7 @@ import {
type ProviderRuntimeEvent,
type ProviderSessionId,
} from "@t3tools/contracts";
-import { Effect, Layer, Queue, Stream } from "effect";
+import { Cause, Effect, Layer, Queue, Stream } from "effect";
import { ProviderService } from "../../provider/Services/ProviderService.ts";
import { OrchestrationEngineService } from "../Services/OrchestrationEngine.ts";
@@ -324,13 +324,27 @@ const make = Effect.gen(function* () {
).pipe(Effect.asVoid);
});
+ const processEventSafely = (event: ProviderRuntimeEvent) =>
+ processEvent(event).pipe(
+ Effect.catchCause((cause) => {
+ if (Cause.hasInterruptsOnly(cause)) {
+ return Effect.failCause(cause);
+ }
+ return Effect.logWarning("provider runtime ingestion failed to process event", {
+ eventId: event.eventId,
+ eventType: event.type,
+ cause: Cause.pretty(cause),
+ });
+ }),
+ );
+
const start: ProviderRuntimeIngestionShape["start"] = Effect.gen(function* () {
const providerEventQueue = yield* Queue.unbounded<ProviderRuntimeEvent>();
yield* Effect.addFinalizer(() => Queue.shutdown(providerEventQueue).pipe(Effect.asVoid));
yield* Effect.forkScoped(
- Effect.forever(Queue.take(providerEventQueue).pipe(Effect.flatMap(processEvent))),
+ Effect.forever(Queue.take(providerEventQueue).pipe(Effect.flatMap(processEventSafely))),
);
yield* Effect.forkScoped(
Stream.runForEach(providerService.streamEvents, (event) =>
diff --git a/apps/server/src/wsServer.test.ts b/apps/server/src/wsServer.test.ts
index 121b75d07..b0057acd6 100644
--- a/apps/server/src/wsServer.test.ts
+++ b/apps/server/src/wsServer.test.ts
@@ -1000,6 +1000,65 @@ describe("WebSocket Server", () => {
expect(response!.error!.message).toContain("Invalid request format");
});
+ it("catches websocket message handler rejections and keeps the socket usable", async () => {
+ const unhandledRejections: unknown[] = [];
+ const onUnhandledRejection = (reason: unknown) => {
+ unhandledRejections.push(reason);
+ };
+ process.on("unhandledRejection", onUnhandledRejection);
+
+ const brokenOpenService: OpenShape = {
+ openBrowser: () => Effect.void,
+ openInEditor: () =>
+ Effect.sync(() => BigInt(1)).pipe(Effect.map((result) => result as unknown as void)),
+ };
+
+ try {
+ server = await createTestServer({ cwd: "/test", open: brokenOpenService });
+ const addr = server.address();
+ const port = typeof addr === "object" && addr !== null ? addr.port : 0;
+
+ const ws = await connectWs(port);
+ connections.push(ws);
+ await waitForMessage(ws);
+
+ ws.send(
+ JSON.stringify({
+ id: "req-broken-open",
+ body: {
+ _tag: WS_METHODS.shellOpenInEditor,
+ cwd: "/tmp",
+ editor: "cursor",
+ },
+ }),
+ );
+
+ await new Promise((resolve) => setTimeout(resolve, 50));
+ expect(unhandledRejections).toHaveLength(0);
+
+ const workspace = makeTempDir("t3code-ws-handler-still-usable-");
+ fs.writeFileSync(path.join(workspace, "file.txt"), "ok\n", "utf8");
+ const response = await sendRequest(ws, WS_METHODS.projectsSearchEntries, {
+ cwd: workspace,
+ query: "file",
+ limit: 5,
+ });
+ expect(response.error).toBeUndefined();
+ expect(response.result).toEqual(
+ expect.objectContaining({
+ entries: expect.arrayContaining([
+ expect.objectContaining({
+ path: "file.txt",
+ kind: "file",
+ }),
+ ]),
+ }),
+ );
+ } finally {
+ process.off("unhandledRejection", onUnhandledRejection);
+ }
+ });
+
it("returns errors for removed projects CRUD methods", async () => {
server = await createTestServer({ cwd: "/test" });
const addr = server.address();
diff --git a/apps/server/src/wsServer.ts b/apps/server/src/wsServer.ts
index 2afceede5..1b0a8ec1e 100644
--- a/apps/server/src/wsServer.ts
+++ b/apps/server/src/wsServer.ts
@@ -542,7 +542,11 @@ export const createServer = Effect.fn(function* (options: ServerOptions = {}): E
ws.send(JSON.stringify(welcome));
ws.on("message", (raw) => {
- void Effect.runPromise(handleMessage(ws, raw));
+ void Effect.runPromise(
+ handleMessage(ws, raw).pipe(
+ Effect.catch((error) => Effect.logError("Error handling message", error)),
+ ),
+ );
});
ws.on("close", () => {