perf(server): read pull request metadata from a lightweight snapshot#30356

The pull-request service now obtains thread metadata and the maximum thread-event sequence through a projection-store snapshot API. The API supports global and thread-targeted reads, with filters for archived and settled threads.

Changes: Pull-request snapshot flow

Layer / File(s)	Summary
Projection-store snapshot apps/server/src/orchestration-v2/ProjectionStore.ts, apps/server/src/orchestration-v2/ProjectionStore.test.ts	The store adds a snapshot query that reads selected thread data and the maximum event sequence in one transaction. It applies thread and unsettled filters, and tests check query scope, transaction grouping, and returned metadata.
Orchestrator and service integration apps/server/src/orchestration-v2/Orchestrator.ts, apps/server/src/orchestration-v2/ThreadPullRequestService.ts, apps/server/src/orchestration-v2/ThreadPullRequestService.test.ts, apps/server/src/orchestration-v2/ProviderTurnControlService.test.ts	The orchestrator exposes the snapshot query, and the pull-request service uses it for global and thread-targeted reads. Tests cover snapshot filtering and the sequence guard in the sync command.

Estimated code review effort: 3 (Moderate) | ~20 minutes

Change: Refactor

Sequence Diagram(s)

mermaid

1	sequenceDiagram
2	participant ThreadPullRequestService
3	participant OrchestratorV2
4	participant ProjectionStoreV2
5	participant ReadOnlySQLiteClient
6	ThreadPullRequestService->>OrchestratorV2: Request pull-request snapshot
7	OrchestratorV2->>ProjectionStoreV2: Delegate snapshot options
8	ProjectionStoreV2->>ReadOnlySQLiteClient: Read selected threads and event sequence in one transaction
9	ReadOnlySQLiteClient-->>ProjectionStoreV2: Return rows and sequence
10	ProjectionStoreV2-->>OrchestratorV2: Return decoded snapshot
11	OrchestratorV2-->>ThreadPullRequestService: Return snapshot and sequence

Possibly related issues

• pingdotgg/t3code#18618: The issue describes full-shell reads for PR discovery; this change uses a filtered metadata snapshot and sequence read.
• pingdotgg/t3code#24895: The issue identifies a slow ThreadPullRequestServiceV2.synchronize shell read; this change replaces the PR-service read with a metadata snapshot.

Possibly related PRs

• pingdotgg/t3code#25728: Adds a related PR-sync snapshot read path that this change extends with a PR-specific API and filters.
• pingdotgg/t3code#25443: Moves PR discovery to thread metadata; this change implements a snapshot API with sequence and thread filters.
• pingdotgg/t3code#13691: Replaces full thread-list reads with targeted snapshot reads for per-thread pull-request checks.

Suggested reviewers: juliusmarminge, macroscopeapp

No actionable issue is established from the supplied evidence; the change is mergeable after normal checks.

Check name	Status	Explanation
Title check	Passed	The title "perf(server): read pull request metadata from a lightweight snapshot" matches the main change. The PR adds a PR-specific snapshot (getPullRequestSnapshot) that reads only thread metadata an…
Description check	Passed	The description follows the template's main sections: Problem, Change, Scope and approval, and Verification. The Problem section explains the slow startup backfill. The Change section describes the ne…
Linked Issues check	Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	Passed	Check skipped because no linked issues were found for this pull request.

Full details: Title check

Explanation

The title "perf(server): read pull request metadata from a lightweight snapshot" matches the main change. The PR adds a PR-specific snapshot (getPullRequestSnapshot) that reads only thread metadata and event sequence, and the PR service uses it for backfill and per-thread refresh. It uses a conventional commit prefix and plain language.

Full details: Description check

Explanation

The description follows the template's main sections: Problem, Change, Scope and approval, and Verification. The Problem section explains the slow startup backfill. The Change section describes the new snapshot and the service updates. The Verification section reports benchmark numbers (1,168 ms to 111 ms), sync command and branch lookup parity, focused test results, and the limits of the measurements. The Scope section explains why this is a focused performance fix. Two gaps remain: it does not link a triaged issue or maintainer approval, and the template asks for approval of direction and scope for changes like this. It also does not clearly state the reason for the performance fix's qualification under the approval rule beyond stating it is focused. The model line (GPT-6 Codex) is present as the template requests.

Resolution

Consider linking the triaged issue or maintainer discussion that approves this performance direction. If none exists, state why this small, focused performance fix qualifies without prior approval.

Generate docstrings

• Commit to this branch
• Create a new PR

Generate unit tests (beta)

• Commit to this branch
• Create a new PR

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 108.5 KiB
• Claude decoded thread snapshot: 108.8 KiB