fix(gmail): drop skip_todo_writeback; rely on platform echo suppression

[KrisBraun]

What

Removes the Gmail connector's local skip_todo_writeback KV guard. It was a per-connector workaround for the write-back echo loop; that suppression now lives in the platform (per-dimension write-provenance on thread_state + echo-filtered dispatch views).

• Deletes the skip_todo_writeback:{id}set (inbound star handler) and the check/clear guard at the top of onThreadToDoFn.
• Keeps the unread:{id} / starred:{id} writes — they double as inbound change-detection baselines and stay until a later inbound rework.
• Adds a regression test: onThreadToDo writes back even when a stale skip_todo_writeback key is present (proving the guard is gone).

Why

Paired with the platform fix in plotday/core (writeback-echo-loop), which stamps connector-caused thread_state writes with provenance so the dispatch views no longer echo a connector's own read/star sync-ins back to it. With that in place, this KV is redundant.

Notes

• Connector package → no changeset (deploys via plot deploy, not npm).
• Merge this before the parent plotday/core PR bumps the submodule pointer.
• Follow-up: outlook-mail has a near-identical skip_todo_writeback (now redundant, safe to remove); slack has a different variant.

🤖 Generated with Claude Code

[KrisBraun]

What

Removes the Gmail connector's local skip_todo_writeback KV guard. It was a per-connector workaround for the write-back echo loop; that suppression now lives in the platform (per-dimension write-provenance on thread_state + echo-filtered dispatch views).

• Deletes the skip_todo_writeback:{id}set (inbound star handler) and the check/clear guard at the top of onThreadToDoFn.
• Keeps the unread:{id} / starred:{id} writes — they double as inbound change-detection baselines and stay until a later inbound rework.
• Adds a regression test: onThreadToDo writes back even when a stale skip_todo_writeback key is present (proving the guard is gone).

Why

Paired with the platform fix in plotday/core (writeback-echo-loop), which stamps connector-caused thread_state writes with provenance so the dispatch views no longer echo a connector's own read/star sync-ins back to it. With that in place, this KV is redundant.

Notes

• Connector package → no changeset (deploys via plot deploy, not npm).
• Merge this before the parent plotday/core PR bumps the submodule pointer.
• Follow-up: outlook-mail has a near-identical skip_todo_writeback (now redundant, safe to remove); slack has a different variant.

🤖 Generated with Claude Code

[KrisBraun]

What

Removes the Gmail connector's local skip_todo_writeback KV guard. It was a per-connector workaround for the write-back echo loop; that suppression now lives in the platform (per-dimension write-provenance on thread_state + echo-filtered dispatch views).

• Deletes the skip_todo_writeback:{id}set (inbound star handler) and the check/clear guard at the top of onThreadToDoFn.
• Keeps the unread:{id} / starred:{id} writes — they double as inbound change-detection baselines and stay until a later inbound rework.
• Adds a regression test: onThreadToDo writes back even when a stale skip_todo_writeback key is present (proving the guard is gone).

Why

Paired with the platform fix in plotday/core (writeback-echo-loop), which stamps connector-caused thread_state writes with provenance so the dispatch views no longer echo a connector's own read/star sync-ins back to it. With that in place, this KV is redundant.

Notes

• Connector package → no changeset (deploys via plot deploy, not npm).
• Merge this before the parent plotday/core PR bumps the submodule pointer.
• Follow-up: outlook-mail has a near-identical skip_todo_writeback (now redundant, safe to remove); slack has a different variant.

🤖 Generated with Claude Code

[KrisBraun]

What

Removes the Gmail connector's local skip_todo_writeback KV guard. It was a per-connector workaround for the write-back echo loop; that suppression now lives in the platform (per-dimension write-provenance on thread_state + echo-filtered dispatch views).

• Deletes the skip_todo_writeback:{id}set (inbound star handler) and the check/clear guard at the top of onThreadToDoFn.
• Keeps the unread:{id} / starred:{id} writes — they double as inbound change-detection baselines and stay until a later inbound rework.
• Adds a regression test: onThreadToDo writes back even when a stale skip_todo_writeback key is present (proving the guard is gone).

Why

Paired with the platform fix in plotday/core (writeback-echo-loop), which stamps connector-caused thread_state writes with provenance so the dispatch views no longer echo a connector's own read/star sync-ins back to it. With that in place, this KV is redundant.

Notes

• Connector package → no changeset (deploys via plot deploy, not npm).
• Merge this before the parent plotday/core PR bumps the submodule pointer.
• Follow-up: outlook-mail has a near-identical skip_todo_writeback (now redundant, safe to remove); slack has a different variant.

🤖 Generated with Claude Code

[KrisBraun]

What

Removes the Gmail connector's local skip_todo_writeback KV guard. It was a per-connector workaround for the write-back echo loop; that suppression now lives in the platform (per-dimension write-provenance on thread_state + echo-filtered dispatch views).

• Deletes the skip_todo_writeback:{id}set (inbound star handler) and the check/clear guard at the top of onThreadToDoFn.
• Keeps the unread:{id} / starred:{id} writes — they double as inbound change-detection baselines and stay until a later inbound rework.
• Adds a regression test: onThreadToDo writes back even when a stale skip_todo_writeback key is present (proving the guard is gone).

Why

Paired with the platform fix in plotday/core (writeback-echo-loop), which stamps connector-caused thread_state writes with provenance so the dispatch views no longer echo a connector's own read/star sync-ins back to it. With that in place, this KV is redundant.

Notes

• Connector package → no changeset (deploys via plot deploy, not npm).
• Merge this before the parent plotday/core PR bumps the submodule pointer.
• Follow-up: outlook-mail has a near-identical skip_todo_writeback (now redundant, safe to remove); slack has a different variant.

🤖 Generated with Claude Code

[KrisBraun]

What

Removes the Gmail connector's local skip_todo_writeback KV guard. It was a per-connector workaround for the write-back echo loop; that suppression now lives in the platform (per-dimension write-provenance on thread_state + echo-filtered dispatch views).

• Deletes the skip_todo_writeback:{id}set (inbound star handler) and the check/clear guard at the top of onThreadToDoFn.
• Keeps the unread:{id} / starred:{id} writes — they double as inbound change-detection baselines and stay until a later inbound rework.
• Adds a regression test: onThreadToDo writes back even when a stale skip_todo_writeback key is present (proving the guard is gone).

Why

Paired with the platform fix in plotday/core (writeback-echo-loop), which stamps connector-caused thread_state writes with provenance so the dispatch views no longer echo a connector's own read/star sync-ins back to it. With that in place, this KV is redundant.

Notes

• Connector package → no changeset (deploys via plot deploy, not npm).
• Merge this before the parent plotday/core PR bumps the submodule pointer.
• Follow-up: outlook-mail has a near-identical skip_todo_writeback (now redundant, safe to remove); slack has a different variant.

🤖 Generated with Claude Code

[KrisBraun]

What

Removes the Gmail connector's local skip_todo_writeback KV guard. It was a per-connector workaround for the write-back echo loop; that suppression now lives in the platform (per-dimension write-provenance on thread_state + echo-filtered dispatch views).

• Deletes the skip_todo_writeback:{id}set (inbound star handler) and the check/clear guard at the top of onThreadToDoFn.
• Keeps the unread:{id} / starred:{id} writes — they double as inbound change-detection baselines and stay until a later inbound rework.
• Adds a regression test: onThreadToDo writes back even when a stale skip_todo_writeback key is present (proving the guard is gone).

Why

Paired with the platform fix in plotday/core (writeback-echo-loop), which stamps connector-caused thread_state writes with provenance so the dispatch views no longer echo a connector's own read/star sync-ins back to it. With that in place, this KV is redundant.

Notes

• Connector package → no changeset (deploys via plot deploy, not npm).
• Merge this before the parent plotday/core PR bumps the submodule pointer.
• Follow-up: outlook-mail has a near-identical skip_todo_writeback (now redundant, safe to remove); slack has a different variant.

🤖 Generated with Claude Code

[KrisBraun]

What

Removes the Gmail connector's local skip_todo_writeback KV guard. It was a per-connector workaround for the write-back echo loop; that suppression now lives in the platform (per-dimension write-provenance on thread_state + echo-filtered dispatch views).

• Deletes the skip_todo_writeback:{id}set (inbound star handler) and the check/clear guard at the top of onThreadToDoFn.
• Keeps the unread:{id} / starred:{id} writes — they double as inbound change-detection baselines and stay until a later inbound rework.
• Adds a regression test: onThreadToDo writes back even when a stale skip_todo_writeback key is present (proving the guard is gone).

Why

Paired with the platform fix in plotday/core (writeback-echo-loop), which stamps connector-caused thread_state writes with provenance so the dispatch views no longer echo a connector's own read/star sync-ins back to it. With that in place, this KV is redundant.

Notes

• Connector package → no changeset (deploys via plot deploy, not npm).
• Merge this before the parent plotday/core PR bumps the submodule pointer.
• Follow-up: outlook-mail has a near-identical skip_todo_writeback (now redundant, safe to remove); slack has a different variant.

🤖 Generated with Claude Code