feat: migrate v1 state into orchestrator v2

[juliusmarminge]

Summary

• converge packaged server, desktop, and SSH runtime state paths back to the production userdata directory
• import v1 materialized thread shells synchronously, then hydrate complete transcripts lazily and in bounded background batches
• hand imported visible history to the first v2 provider turn without trying to migrate v1 provider sessions or checkpoints
• add migration 42 to track idempotent shell and transcript import progress

Migration behavior

• startup work is proportional to thread shells plus at most two preview messages per thread; it does not replay the raw v1 event log
• full transcripts are read from projection_thread_messages after command readiness and on first open/dispatch
• deterministic IDs and a migration manifest make imports idempotent and retryable after interruption
• migrated history supports normal viewing and continuation; v1 runs, checkpoints, forks, and tool activity are intentionally not reconstructed, while newly created v2 runs keep the full v2 feature set

Verification

• 83 focused tests across 14 changed-path test files, plus a final 19-test service regression pass after restacking
• targeted typechecks for server, contracts, desktop, and SSH packages
• targeted lint, format, and diff checks
• isolated browser verification that a seeded v1 thread is imported, rendered, and remains interactive

Stacked on #2829. The released migration-order repair now lives in the parent PR.

---

Note

High Risk
Touches startup ordering, SQLite migration, on-disk state paths, and command/WebSocket paths where transcript hydration failures block reads and dispatch; incorrect import or path rename could strand or duplicate user data.

Overview
Migrates legacy v1 conversation state into orchestration v2 and drops the userdata-v2 / t3code-v2 naming in favor of production userdata and t3code (desktop, server CLI, SSH remote launch).

A new LegacyV1ThreadImporter runs at startup (reconcileShells) to emit v2 events for threads that exist only in v1 projections, with lightweight shell previews (latest user + latest message). Full transcripts hydrate lazily via ensureTranscript on dispatch/projection/WebSocket subscribe, plus a background importPendingTranscripts pass after the server accepts commands. Progress is tracked in migration 043 (orchestration_v2_legacy_imports).

Imported threads get historyOrigin: v1_import. prepareLegacyImport builds a truncated manual_context handoff so the first native v2 turn (and provider-switch recovery after a failed run) can see imported user/assistant text without migrating v1 provider sessions. isTurnItemAtOrBeforeRun treats runless imported items as visible for forks and provider switches.

ThreadManagementService hydrates legacy transcripts before dispatch and projection reads; hydration failures surface as dispatch/projection errors.

^{Reviewed by Cursor Bugbot for commit 67ab204. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Migrate v1 thread state into orchestrator v2 with incremental import and context handoff

• Introduces LegacyV1ThreadImporter to migrate v1 threads into the v2 orchestration system in two phases: reconcileShells (run at startup) creates thread shells with preview events, and importPendingTranscripts (background fiber) hydrates full transcripts idempotently.
• Adds prepareLegacyImport to ContextHandoffServiceV2 to generate manual_context handoffs carrying a text summary of imported history, capped at 32,000 characters with word-boundary truncation.
• Integrates legacy transcript hydration into ThreadManagementService so projections and dispatches hydrate v1 threads before execution, surfacing failures as OrchestratorProjectionError or OrchestratorDispatchError.
• Adds historyOrigin ('native' | 'v1_import') to OrchestrationV2AppThread and OrchestrationV2ThreadShell contracts, and propagates it through projection and shell utilities.
• Adds DB migration 043 (orchestration_v2_legacy_imports) to track per-thread import progress.
• Renames non-dev state directory from userdata-v2 to userdata across server, desktop, and SSH tunnel paths.
• Risk: the userdata-v2 → userdata rename is a breaking path change; existing non-dev installations must migrate their state directory or data will not be found.

^{Macroscope summarized 67ab204.}

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 175c41d2-e7ea-42d0-bd56-40cc466ee70a

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch codex/v1-v2-state-migration

---

_{Comment @coderabbitai help to get the list of available commands.}