feat: local to cloud handoff

[jonathanlab]

Now let's do it the other way around as well, local to cloud

MERGE THIS PR IN THE MONOREPO BEFORE MERGING THIS ONE: PostHog/posthog#54540

[jonathanlab]

• feat: remove tree snapshots, use checkpoints, fix handoff bugs #1665
• feat: add confirmation ui for handoff #1647
• feat: local to cloud handoff #1618 👈 (View in Graphite)
• feat: use checkpoints for handoff #1607
• feat: cloud to local handoff #1522
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.

[jonathanlab]

Merge activity

• Apr 23, 2:04 PM UTC: A user started a stack merge that includes this pull request via Graphite.
• Apr 23, 2:17 PM UTC: Graphite rebased this pull request as part of a merge.
• Apr 23, 2:19 PM UTC: Graphite couldn't merge this PR because it was not satisfying all requirements (Failed CI: 'build').
• Apr 23, 3:02 PM UTC: @jonathanlab merged this pull request with Graphite.

[greptile-apps]

Comments Outside Diff (2)

1. apps/code/src/renderer/hooks/useFeatureFlag.ts, line 19-27 (link)

   All feature flags enabled in dev, not just the cloud-handoff flag

   IS_DEV || ... short-circuits for every call to useFeatureFlag, enabling every flag across the whole app when running locally — not just CLOUD_HANDOFF_FLAG. Any other flag that is intentionally gated (e.g. a half-finished feature, a risky rollout) will silently become active in development. Consider passing the override at the call site, or maintaining a specific allowlist of dev-overridden flags rather than enabling all of them unconditionally.

   Prompt To Fix With AI

   This is a comment left during a code review.
   Path: apps/code/src/renderer/hooks/useFeatureFlag.ts
   Line: 19-27
   
   Comment:
   **All feature flags enabled in dev, not just the cloud-handoff flag**
   
   `IS_DEV || ...` short-circuits for every call to `useFeatureFlag`, enabling every flag across the whole app when running locally — not just `CLOUD_HANDOFF_FLAG`. Any other flag that is intentionally gated (e.g. a half-finished feature, a risky rollout) will silently become active in development. Consider passing the override at the call site, or maintaining a specific allowlist of dev-overridden flags rather than enabling all of them unconditionally.
   
   How can I resolve this? If you propose a fix, please make it concise.

2. apps/code/src/main/services/handoff/handoff-to-cloud-saga.ts, line 354-356 (link)

   Side-effectful killSession wrapped in readOnlyStep

   readOnlyStep signals "no rollback needed", but terminating the local agent is an irreversible side effect. If the subsequent update_workspace step fails and rolls back (setting the mode back to local), the local agent is already dead, leaving the workspace in an inconsistent state where the mode says local but no agent is running. Consider making this a normal step with a no-op or reconnect attempt as the rollback, or at minimum swapping the order so update_workspace runs before the session is killed.

   Prompt To Fix With AI

   This is a comment left during a code review.
   Path: apps/code/src/main/services/handoff/handoff-to-cloud-saga.ts
   Line: 354-356
   
   Comment:
   **Side-effectful `killSession` wrapped in `readOnlyStep`**
   
   `readOnlyStep` signals "no rollback needed", but terminating the local agent is an irreversible side effect. If the subsequent `update_workspace` step fails and rolls back (setting the mode back to `local`), the local agent is already dead, leaving the workspace in an inconsistent state where the mode says `local` but no agent is running. Consider making this a normal `step` with a no-op or reconnect attempt as the rollback, or at minimum swapping the order so `update_workspace` runs before the session is killed.
   
   How can I resolve this? If you propose a fix, please make it concise.

Prompt To Fix All With AI

This is a comment left during a code review.
Path: apps/code/src/main/services/handoff/service.ts
Line: 337

Comment:
**Incorrect return value inflates `processedLineCount`**

`flushLocalLogs` returns `lines.length` — the total count of non-empty lines in the file, including the `seed_boundary` marker and any seeded cloud entries from a previous run. Only `entries` (lines after the boundary that parse as valid JSON) are actually appended to the cloud log. When the return value flows through `flushedLogEntryCount` → `logEntryCount` → `processedLineCount` on the session, the cloud watcher starts reading the cloud log offset by those extra lines and will skip valid new entries that were just flushed.

```suggestion
        return entries.length;
```

How can I resolve this? If you propose a fix, please make it concise.

---

This is a comment left during a code review.
Path: apps/code/src/renderer/hooks/useFeatureFlag.ts
Line: 19-27

Comment:
**All feature flags enabled in dev, not just the cloud-handoff flag**

`IS_DEV || ...` short-circuits for every call to `useFeatureFlag`, enabling every flag across the whole app when running locally — not just `CLOUD_HANDOFF_FLAG`. Any other flag that is intentionally gated (e.g. a half-finished feature, a risky rollout) will silently become active in development. Consider passing the override at the call site, or maintaining a specific allowlist of dev-overridden flags rather than enabling all of them unconditionally.

How can I resolve this? If you propose a fix, please make it concise.

---

This is a comment left during a code review.
Path: apps/code/src/main/services/handoff/handoff-to-cloud-saga.ts
Line: 354-356

Comment:
**Side-effectful `killSession` wrapped in `readOnlyStep`**

`readOnlyStep` signals "no rollback needed", but terminating the local agent is an irreversible side effect. If the subsequent `update_workspace` step fails and rolls back (setting the mode back to `local`), the local agent is already dead, leaving the workspace in an inconsistent state where the mode says `local` but no agent is running. Consider making this a normal `step` with a no-op or reconnect attempt as the rollback, or at minimum swapping the order so `update_workspace` runs before the session is killed.

How can I resolve this? If you propose a fix, please make it concise.

_{Reviews (1): Last reviewed commit: "chore: trigger ci" | Re-trigger Greptile}