Codex review: needs maintainer review before merge. Reviewed July 11, 2026, 6:53 PM ET / 22:53 UTC.

Summary
The PR migrates legacy Microsoft Teams conversation references under embedded canonical conversation IDs, verifies visibility before archival, retains the source after failed or incomplete migration, and adds production SQLite failure/retry coverage.

PR surface: Source +26, Tests +119. Total +145 across 3 files.

Reproducibility: yes. from source and supplied exact-head evidence. The production-store regression and upgrade transcript establish the failure, retained source, retry, canonical lookup, and archival sequence, although this read-only review did not independently execute the PR.

Review metrics: 1 noteworthy metric.

• Persistent identity: 1 migration key rule changed. Legacy references now use the canonical conversation identity expected by proactive runtime lookup.

Stored data model
Persistent data-model change detected: migration/backfill/repair: extensions/msteams/doctor-contract-api.test.ts, migration/backfill/repair: extensions/msteams/doctor-contract-api.ts, serialized state: extensions/msteams/doctor-contract-api.test.ts, unknown-data-model-change: extensions/msteams/doctor-contract-api.test.ts, unknown-data-model-change: extensions/msteams/doctor-contract-api.ts. Confirm migration or upgrade compatibility proof before merge.

Root-cause cluster
Relationship: fixed_by_candidate
Canonical: #94939
Summary: The linked issue is the canonical migration regression; this PR is its active candidate fix, while the earlier contribution was replaced without merge.

Members:

• canonical: [Bug]: 6.x state migration leaves channel conversation-store SQLite empty (0 bytes) — orphans references, breaks proactive (Bot Framework) sends (MS Teams) #94939 - It reports the same archived legacy-reference and proactive Microsoft Teams send failure addressed by this patch.
• superseded: fix(msteams): proactive sends fail after conversation migration #98862 - It is the closed-unmerged predecessor explicitly replaced by this writable-head PR.

Proposal only: this assessment does not dispatch repair, suppress jobs, mutate sibling items, close, or merge anything.

Merge readiness
Overall: 🐚 platinum hermit
Proof: 🦞 diamond lobster
Patch quality: 🐚 platinum hermit
Result: ready for maintainer review.

Overall follows the weaker of proof and patch quality, so missing proof can cap an otherwise strong patch.

Rank-up moves:

• none.

Risk before merge

• [P1] The PR is mergeable but behind current main, so the final merged head should preserve newer main changes and repeat the exact-head compatibility and migration gates.

Maintainer options:

1. Land with current-base gates (recommended)
   Merge after the final current-base head preserves the reviewed migration behavior and all required checks pass.

Next step before merge

• No automated repair is needed; the remaining action is routine maintainer landing review on the final current-base head.

Security
Cleared: The patch introduces no dependency, workflow, permission, secret, package-resolution, artifact-download, or new code-execution surface.

Exact-head proof update for 9ab0693285ef1bd97c8756444a7adb18f6d591df:

• GitHub CI is fully green on this exact head: zero pending and zero failing checks. Relevant passing lanes include check-lint, check-additional-extension-bundled, check-additional-extension-channels, Real behavior proof, QA smoke profiles, Node compact shards, type checks, security checks, and boundary checks.
• Local validation on the same head before push:
  • node scripts/run-vitest.mjs extensions/msteams/doctor-contract-api.test.ts: 7 tests passed
  • node scripts/run-vitest.mjs extensions/msteams: 75 files / 1099 tests passed
  • pnpm run lint:extensions:bundled: passed
  • pnpm run lint:extensions:channels: passed; no extension channel files found
  • pnpm lint --threads=8: passed
  • git diff --check: passed
• Scope clarification for the proof gap: this PR should be reviewed as a migration-safety/preservation guard, not as a full canonical shared-state repair. It does not claim that the early-return path itself restores runtime-visible conversation references. The intended behavior is narrower: when a non-empty legacy msteams-conversations.json has retained rows attempted but the migration observes imported === 0 and present === 0, the doctor emits a warning and leaves the recoverable source in place instead of silently archiving it.
• Compatibility/message-delivery risk mitigation: successful verified migrations keep the existing archive behavior; empty legacy sources keep the existing behavior; only the unverifiable non-empty case preserves recoverable legacy data so an operator or later corrected migration still has the conversation references needed for recovery.
• Redacted runtime proof remains in the PR body: patched bundle installed, migration guard present in patched bundle and absent from baseline bundle, gateway/listener running, msteamsConversationReferences: 20, sent-message records increased 6 -> 7 after a real proactive Teams send, and the latest personal conversation timestamp updated. Conversation IDs, tenant/user IDs, message bodies, hostnames, and local usernames are redacted.

Given the exact-head green CI and the narrowed claim above, please re-review this as the narrower preservation/safety fix rather than a complete canonical proactive-send repair.

@clawsweeper re-review

🦞🧹
ClawSweeper re-review requested.

I asked ClawSweeper to review this item again.
Action: item re-review queued (workflow sweep.yml, event repository_dispatch).
Result: the existing ClawSweeper review comment will be edited in place when the review finishes.

Re-review progress:

• State: Complete
• Detail: The targeted re-review published an exact durable review comment and queued an executing target-wide serialized router scan.
• Run: https://github.com/openclaw/clawsweeper/actions/runs/29169176995
• Updated: 2026-07-11T21:47:22.268Z

Updated at exact head ed7b433 with the real production-store regression ClawSweeper requested.

Evidence covered by the new regression in extensions/msteams/doctor-contract-api.test.ts:

• Uses the production doctor migration context and SQLite-backed plugin keyed store, not a fake unsupported store state.
• Creates a legacy msteams-conversations.json entry keyed by a legacy/noncanonical JSON key while the Bot Framework reference contains the canonical conversation.id used for proactive sends.
• Pre-fills plugin state, sets the production store test row limit to 1, and verifies the first migration rejects with PLUGIN_STATE_LIMIT_EXCEEDED while leaving msteams-conversations.json in place and not creating .migrated.
• Removes the row limit, resets the store, retries the same migration, and verifies no warnings, Migrated 1, source archive creation, canonical lookup by reference.conversation.id succeeds, and lookup by the legacy JSON key remains undefined.

Production change: conversation migration now resolves the stored Bot Framework reference conversation.id before falling back to the legacy JSON key, so archived migrations leave rows visible at the same canonical key the Teams proactive send path uses.

Validation on this head:

• node scripts/run-vitest.mjs extensions/msteams/doctor-contract-api.test.ts
• node scripts/run-vitest.mjs extensions/msteams
• pnpm run lint:extensions:bundled
• pnpm run lint:extensions:channels
• git diff --check
• pnpm lint --threads=8
• pnpm plugin-sdk:check-exports
• gh pr checks 104691 --repo openclaw/openclaw: zero failed, zero pending; Real behavior proof passed.

@clawsweeper re-review

🦞🧹
ClawSweeper re-review requested.

I asked ClawSweeper to review this item again.
Action: item re-review queued (workflow sweep.yml, event repository_dispatch).
Result: the existing ClawSweeper review comment will be edited in place when the review finishes.

Re-review progress:

• State: Complete
• Detail: The targeted re-review published an exact durable review comment and queued an executing target-wide serialized router scan.
• Run: https://github.com/openclaw/clawsweeper/actions/runs/29171066710
• Updated: 2026-07-11T22:55:00.636Z

Exact-head redacted upgrade-path proof for ed7b43389155f9f6c0e3919b8079f532797502d9.

This proof uses a disposable local OpenClaw state home under /var/.../T, with no live gateway, no Dale state, and no real tenant/user/conversation IDs. To avoid stale ignored dist/extensions artifacts in the checkout, the runner imports the exact PR-head source contract at extensions/msteams/doctor-contract-api.ts and exercises it with the same production plugin-state SQLite keyed-store context used by doctor migrations (pluginId: msteams, namespace conversations).

The fixture is an authentic legacy msteams-conversations.json shape: one retained row whose outer JSON key is legacy-user-key, while reference.conversation.id is the canonical Teams conversation ID used for proactive lookup.

Redacted proof transcript:

{
  "head": "ed7b43389155f9f6c0e3919b8079f532797502d9",
  "proofMode": "exact PR-head source msteams doctor contract + production plugin-state SQLite context",
  "disposableStateDir": "<tmp-proof-root>/state-dir",
  "fixture": {
    "legacyJsonKey": "legacy-user-key",
    "embeddedConversationId": "19:[email protected]",
    "retainedConversationEntries": 1
  },
  "afterFailure": {
    "failureCode": "PLUGIN_STATE_LIMIT_EXCEEDED",
    "failureMessage": "Plugin state for msteams exceeds the 1 live row limit.",
    "legacyExists": true,
    "legacySizeBytes": 552,
    "archiveExists": false,
    "sqliteExists": true
  },
  "afterRetry": {
    "changes": [
      "Migrated 1 Microsoft Teams conversation entry -> plugin state",
      "Archived Microsoft Teams conversation legacy source -> <tmp-proof-root>/state-dir/msteams-conversations.json.migrated"
    ],
    "warningCount": 0,
    "legacyExists": false,
    "archiveExists": true,
    "archiveSizeBytes": 552,
    "sqliteExists": true
  },
  "canonicalVisibility": {
    "canonicalKeyPrefix": "36fb5f4161c2",
    "legacyKeyPrefix": "8987278c02e0",
    "pluginStateLookupByEmbeddedConversationIdKey": true,
    "pluginStateLookupByLegacyJsonKey": false,
    "storedConversationId": "19:[email protected]",
    "storedUserId": "29:redacted-user"
  },
  "sqlite": {
    "path": "<tmp-proof-root>/state-dir/state/openclaw.sqlite",
    "msteamsNamespaceCounts": [
      {
        "plugin_id": "msteams",
        "namespace": "conversations",
        "rows": 1
      },
      {
        "plugin_id": "msteams",
        "namespace": "proof-sibling",
        "rows": 1
      }
    ],
    "conversationKeyRows": [
      {
        "namespace": "conversations",
        "keyPrefix": "36fb5f4161c2",
        "matchesCanonicalKey": true,
        "matchesLegacyJsonKey": false
      }
    ]
  }
}

What this demonstrates:

• Real production SQLite row-limit failure path: PLUGIN_STATE_LIMIT_EXCEEDED.
• Failed migration leaves the populated legacy source in place and does not create .migrated.
• Retry with a healthy store imports one Microsoft Teams conversation entry.
• Retry archives the source only after successful import.
• Shared plugin-state SQLite contains one msteams/conversations row.
• The stored row matches the canonical embedded reference.conversation.id key, not the legacy outer JSON key.

Exact-head CI remains green: gh pr checks 104691 --repo openclaw/openclaw exits 0 with zero failed and zero pending checks.

@clawsweeper re-review

This pull request has been automatically marked as stale due to inactivity.
Please add updates or it will be closed.

Active maintenance update: this Microsoft Teams conversation-migration PR is not abandoned. The published PR head still matches its fork branch, and I am tracking the current integration state. The exact PR ref currently has unrelated history from upstream/main in a local merge-tree check, so I am preserving the published branch rather than force-pushing or synthesizing history. I am continuing to investigate the focused recovery path.