fix(workflow-orchestration): correct resume and failure-branch routing in WorkflowRunner

- Resume no longer infinite-loops when a crash occurs between
  append_step_result() and update_run() for a BranchStep: the recorded
  branch target is now read from the persisted result instead of
  recomputed as None.
- Resume no longer silently discards a recorded step failure that
  happened right before the crash; the run correctly ends failed.
- A failed step's own next_step_id (e.g. routing to a BranchStep that
  evaluates step_result_success) is now honored instead of
  unconditionally forcing the run to failed, per spec.

openspec: workflow-orchestration capability, archived change workflow-orchestration-runtime

tests/test_workflow_runner.py

@@ -14,6 +14,7 @@ from workflow.models import (
     SkillInvocationStep,
     WaitForConditionStep,
     WorkflowDefinition,
+    WorkflowStepResult,
 )
 from workflow.runner import WorkflowRunner
 from workflow.store import WorkflowStore
@@ -266,6 +267,145 @@ def test_workflow_runner_resume_skips_completed_steps_after_restart(tmp_path) ->
     assert calls == ["one", "two", "three"]
 
 
+def test_workflow_runner_resume_after_branch_crash_advances_to_recorded_target(
+    tmp_path,
+) -> None:
+    """Simulates a crash between append_step_result() and update_run() for a
+    BranchStep: the step's result is persisted but the checkpoint never
+    happened, so current_step_id still points at the branch step itself.
+    Resuming must recompute the next step from the *recorded* branch
+    target instead of losing it (and must not loop forever re-processing
+    the branch step).
+    """
+    calls: list[str] = []
+    store = _store(tmp_path)
+    definition = WorkflowDefinition(
+        name="branch-resume",
+        entry_step_id="branch",
+        steps=[
+            BranchStep(
+                "branch",
+                ConditionSpec("world_variable_equals", {"name": "ready", "value": True}),
+                on_true="true-step",
+                on_false="false-step",
+            ),
+            PlannedGoalStep("false-step", "false path"),
+            PlannedGoalStep("true-step", "true path"),
+        ],
+    )
+    store.save_definition(definition)
+    run = store.create_run(definition.id, {"ready": True}, device_id="phone")
+    store.append_step_result(
+        run.id,
+        WorkflowStepResult(
+            step_id="branch",
+            kind="branch",
+            success=True,
+            detail={"condition_result": True, "next_step_id": "true-step"},
+        ),
+    )
+
+    runner = WorkflowRunner(
+        store,
+        task_runner_factory=lambda: FakeTaskRunner(calls),
+    )
+    resumed = runner.resume(run.id)
+
+    assert resumed.status == "completed"
+    assert calls == ["true path"]
+    assert [result.step_id for result in resumed.step_results] == [
+        "branch",
+        "true-step",
+    ]
+
+
+def test_workflow_runner_resume_after_failed_step_crash_ends_failed(tmp_path) -> None:
+    """Simulates a crash between append_step_result(success=False) and
+    update_run(): the failed result is persisted but current_step_id still
+    points at the failed step. Resuming must recognize the recorded
+    failure and end the run as failed rather than silently continuing (or
+    looping forever re-processing the same step).
+    """
+    store = _store(tmp_path)
+    definition = WorkflowDefinition(
+        name="fail-resume",
+        entry_step_id="only",
+        steps=[PlannedGoalStep("only", "risky")],
+    )
+    store.save_definition(definition)
+    run = store.create_run(definition.id, {}, device_id="phone")
+    store.append_step_result(
+        run.id,
+        WorkflowStepResult(
+            step_id="only",
+            kind="planned_goal",
+            success=False,
+            detail={"task_status": "failed", "failure_reason": "boom"},
+        ),
+    )
+
+    runner = WorkflowRunner(store)
+    resumed = runner.resume(run.id)
+
+    assert resumed.status == "failed"
+    assert resumed.current_step_id is None
+
+
+def test_workflow_runner_failed_step_routes_to_recovery_branch(tmp_path) -> None:
+    """A planned-goal step that fails still routes to its declared
+    next_step_id (a branch step evaluating step_result_success). Live
+    execution must reach and execute the on_false recovery path rather
+    than immediately terminating the run as failed.
+    """
+    calls: list[str] = []
+
+    class ScriptedTaskRunner:
+        def __init__(self, calls: list[str], failing_goals: set[str]) -> None:
+            self.calls = calls
+            self.failing_goals = failing_goals
+
+        def run(self, task: Task) -> Task:
+            self.calls.append(task.goal)
+            if task.goal in self.failing_goals:
+                task.status = "failed"  # type: ignore[assignment]
+                task.failure_reason = "boom"
+            else:
+                task.status = "completed"  # type: ignore[assignment]
+                task.failure_reason = None
+            return task
+
+    definition = WorkflowDefinition(
+        name="failure-recovery",
+        entry_step_id="goal",
+        steps=[
+            PlannedGoalStep("goal", "risky goal", next_step_id="check"),
+            BranchStep(
+                "check",
+                ConditionSpec("step_result_success", {"step_id": "goal"}),
+                on_true="happy",
+                on_false="recover",
+            ),
+            PlannedGoalStep("happy", "happy path"),
+            PlannedGoalStep("recover", "recover path"),
+        ],
+    )
+    runner = WorkflowRunner(
+        _store(tmp_path),
+        task_runner_factory=lambda: ScriptedTaskRunner(calls, {"risky goal"}),
+    )
+
+    run = runner.run(definition, "phone")
+
+    assert run.status == "completed"
+    assert calls == ["risky goal", "recover path"]
+    assert [result.step_id for result in run.step_results] == [
+        "goal",
+        "check",
+        "recover",
+    ]
+    assert run.step_results[0].success is False
+
+
 def test_workflow_run_persists_variables_and_ordered_results(tmp_path) -> None:
     store = _store(tmp_path)
     definition = WorkflowDefinition(

workflow/runner.py

@@ -98,19 +98,21 @@ class WorkflowRunner:
             if self.step_limit is not None and executed >= self.step_limit:
                 return run
             step = definition.step_by_id(run.current_step_id)
-            if _already_recorded(run, step.step_id):
-                next_step_id = self._next_step_id(definition, step, None)
-                run = self._checkpoint(run, next_step_id, "running")
+            recorded = _recorded_result(run, step.step_id)
+            if recorded is not None:
+                branch_next_step_id = None
+                if isinstance(step, BranchStep):
+                    branch_next_step_id = recorded.detail.get("next_step_id")
+                next_status, next_step_id = self._resolve_outcome(
+                    definition, step, recorded.success, branch_next_step_id
+                )
+                run = self._checkpoint(run, next_step_id, next_status)
                 continue
 
             result, branch_next_step_id = self._execute_step(definition, run, step)
-            next_status = "running"
-            next_step_id = self._next_step_id(definition, step, branch_next_step_id)
-            if not result.success:
-                next_status = "failed"
-                next_step_id = None
-            elif next_step_id is None:
-                next_status = "completed"
+            next_status, next_step_id = self._resolve_outcome(
+                definition, step, result.success, branch_next_step_id
+            )
 
             self.store.append_step_result(run.id, result)
             run = self._checkpoint(
@@ -121,6 +123,25 @@ class WorkflowRunner:
             executed += 1
         return run
 
+    def _resolve_outcome(
+        self,
+        definition: WorkflowDefinition,
+        step: WorkflowStep,
+        success: bool,
+        branch_next_step_id: str | None,
+    ) -> tuple[str, str | None]:
+        """Compute the (status, next_step_id) for a step's outcome.
+
+        Shared by live execution and resume so a failed step's routing
+        (e.g. to a branch step evaluating the failure) can never drift
+        between the two code paths. The run only becomes terminal when
+        there is truly no next step to route to.
+        """
+        next_step_id = self._next_step_id(definition, step, branch_next_step_id)
+        if next_step_id is None:
+            return ("completed" if success else "failed"), None
+        return "running", next_step_id
+
     def _checkpoint(
         self,
         run: WorkflowRun,
@@ -317,5 +338,8 @@ class WorkflowRunner:
         return definition.next_step_id_after(step.step_id)
 
 
-def _already_recorded(run: WorkflowRun, step_id: str) -> bool:
-    return any(result.step_id == step_id for result in run.step_results)
+def _recorded_result(run: WorkflowRun, step_id: str) -> WorkflowStepResult | None:
+    for result in reversed(run.step_results):
+        if result.step_id == step_id:
+            return result
+    return None