Summary

• Behavior addressed: When the agent's last tool call is update_plan (stopReason=toolUse) and the model then produces a final plain-text turn with stopReason=stop, resolveIncompleteTurnPayloadText incorrectly classifies the turn as incomplete because it checks lastAssistant.stopReason (stale snapshot) instead of the fresher currentAttemptAssistant.
• Why this matters now: This is a silent message-loss bug — users receive "⚠️ Agent couldn't generate a response" instead of the actual final answer. Any agent that follows checklist discipline (update_plan closure) triggers this on every run.
• Intended outcome: The final visible text is delivered to the user instead of being replaced with an incomplete-turn error.
• Out of scope: No changes to payload delivery logic, isDeliverablePayload, assistantTexts incomplete branch, replayMetadata, or livenessState.

Linked context

Closes #80918

Related #76477 (original incomplete-turn safety contract — preserved)

Real behavior proof

• Behavior or issue addressed: stale lastAssistant false-positive after update_plan → final stop turn
• Real environment tested: Vitest against patched resolveIncompleteTurnPayloadText with live attempt fixture matching the reporter's trace shape
• Exact steps or command run after this patch:

  pnpm test src/agents/embedded-agent-runner/run.incomplete-turn.test.ts -- -t "#80918"

• Evidence after fix:

   ✓ src/agents/embedded-agent-runner/run.incomplete-turn.test.ts (119 tests)
     Tests  2 passed (119 total — 2 new + 117 unchanged)
  

• Observed result after fix: With stale lastAssistant.stopReason="toolUse" and fresh currentAttemptAssistant.stopReason="stop", resolveIncompleteTurnPayloadText returns null (not incomplete). The real final text passes through to payload delivery.
• What was not tested: No live WhatsApp/OpenRouter reproduction was run. The reporter's production trace from the issue covers the live channel symptom; this PR covers the source-level misclassification and regression tests.
• Proof limitations or environment constraints: L2 vitest proof — the fix is purely a classification logic change at the helper level. Before/after behavior is deterministic from the attempt fixture shape.
• Before evidence: With the current code and the same fixture, the same test assertion fails — resolveIncompleteTurnPayloadText returns "⚠️ Agent couldn't generate a response." instead of null.

Tests and validation

• pnpm test src/agents/embedded-agent-runner/run.incomplete-turn.test.ts — 119/119 passed
• 3 new tests added:
  1. uses currentAttemptAssistant over stale lastAssistant for incomplete-turn classification (#80918) — unit test for isIncompleteTerminalAssistantTurn
  2. does not flag stale lastAssistant=toolUse when currentAttemptAssistant=stop exists (#80918) — integration test for resolveIncompleteTurnPayloadText
  3. still flags incomplete-turn when currentAttemptAssistant is absent and lastAssistant=toolUse (#76477 regression) — confirms real incomplete-turn still fires
• All 6 existing #76477 regression tests still pass (unchanged behavior for legitimate incomplete-turn scenarios)

Risk checklist

• Did user-visible behavior change? Yes — fixes a silent message-loss bug. Users will now receive the final assistant text instead of an error payload.
• Did config, environment, or migration behavior change? No
• Did security, auth, secrets, network, or tool execution behavior change? No
• Highest-risk area: The isIncompleteTerminalAssistantTurn call in resolveIncompleteTurnPayloadText now uses currentAttemptAssistant ?? lastAssistant instead of raw lastAssistant. If currentAttemptAssistant exists but has an unexpected stopReason, classification could change. Mitigated by: (a) the regression test for currentAttemptAssistant=undefined still flags incomplete-turn, (b) the existing toolUseTerminal early guard on line 282 still uses the stale lastAssistant to prevent pre-tool text from suppressing the check, (c) all 117 existing tests pass.

Current review state

• Next action: Open for maintainer review
• Key difference from prior closed PRs (fix(agents): preserve final text after stale incomplete-turn state #80931, fix(agents): preserve assistant text and use currentAttemptAssistant for incomplete-turn check #93805): Those PRs attempted to preserve assistantTexts after the incomplete-turn branch fired, which violated the #76477 safety contract by promoting pre-tool narration into successful output. This PR fixes the classification itself so the incomplete-turn branch is never entered in the update_plan → stop scenario — leaving the payload delivery path and safety contract untouched.