code review

docs/superpowers/plans/2026-06-01-wf-api-slice-4b-draft-service.md

@@ -397,7 +397,7 @@ _local_path_payload
 _state_path_payload
 ```
 
-Keep `_escape_json_pointer` if `set_draft_route` delegation still needs no local use. Remove it only if no remaining references exist.
+Remove `_escape_json_pointer` only if no remaining references exist in `handlers.py`.
 
 Do not remove from `handlers.py` yet unless `rg` proves there are no remaining
 callers:

docs/superpowers/research/2026-06-01-wf-api-extraction-map.md

@@ -409,6 +409,7 @@ src/wf_cli/
 - `WfMcpService.__post_init__` creates default `FileWorkflowArtifactStore`, `FileDraftWorkspaceStore`, `FileRunStore` if not provided. These are protocol-neutral stores.
 - The stores are created from `_store_root(self.store)` which uses the MCP `Store` root.
 - After extraction, store creation should be the caller's responsibility (config-driven), not `WfMcpService`'s.
+- Next extraction slice must define the store initialization boundary: which caller constructs `FileWorkflowArtifactStore`, `FileDraftWorkspaceStore`, and `FileRunStore`; how config/API callers inject protocol-neutral store implementations; and which tests must explicitly construct stores instead of relying on `WfMcpService.__post_init__`.
 
 ### Naming Confusion
 
@@ -446,6 +447,7 @@ src/wf_cli/
 8. **Run full test suite** — all existing tests must pass with import-only changes.
 9. **Update `CliContext`** to use `WorkflowApi` instead of `WorkflowSurfaceHandlers`.
 10. **Add integration test** for `WorkflowApi` with a mock backend (not `WfMcpService`).
+11. **Define store ownership** before extracting service construction: remove implicit workflow-store creation from `WfMcpService.__post_init__`, add config/API hooks for injected stores, and update tests to pass stores explicitly.
 
 ### Dependency Graph After Extraction
 

docs/superpowers/research/2026-06-01-wf-api-extraction-mapping-prompt.md

@@ -15,7 +15,7 @@ The current suspicion:
 Write your findings to:
 
 ```text
-random shit/wf_api_extraction_map.md
+docs/superpowers/research/2026-06-01-wf-api-extraction-map.md
 ```
 
 Keep it structured and link-heavy. Do not paste huge code blocks. Use file paths, symbol names, and short notes.

skills/wf-workflow/references/capabilities-and-wrappers.md

@@ -50,3 +50,12 @@ that decides what content types are acceptable.
 
 If a result exposes a convenience `text` field, inspect the capability schema
 and wrapper hints before using it. Do not assume every content block is text.
+
+Incorrect:
+
+```json
+{"source": {"root": "local", "parts": ["content"]}, "target": {"root": "state", "parts": ["summary"]}}
+```
+
+Correct: filter `content` to text blocks, extract each `text`, then combine or
+select the value before writing it to a string state field.

src/wf_api/artifacts.py

@@ -74,7 +74,7 @@ class WorkflowArtifactApi:
                 query=query,
             )
         ]
-        entries.sort(key=lambda entry: str(entry["name"]))
+        entries.sort(key=lambda entry: str(entry.get("name", "")))
         return paged_list_payload("nodes", entries, cursor=cursor, limit=limit)
 
     async def save_artifact(self, artifact: dict[str, Any]) -> dict[str, Any]:

src/wf_api/next_actions.py

@@ -306,6 +306,7 @@ def _bounded_trace_example(
     trace_count: int,
 ) -> NextActionPatchExample:
     """Return a safe read_run_trace request; never suggest full trace reads."""
+    limit = max(1, min(25, trace_count))
     return NextActionPatchExample(
         description=(
             "Read a bounded debug trace slice. Increase start/limit only when needed."
@@ -315,7 +316,7 @@ def _bounded_trace_example(
             "run_id": run_id,
             "trace_range": {
                 "start": 0,
-                "limit": 25,
+                "limit": limit,
             },
         },
     )

src/wf_api/refs.py

@@ -2,11 +2,16 @@ from __future__ import annotations
 
 from typing import Any, TypeAlias
 
+from pydantic import TypeAdapter, ValidationError
+
 from wf_artifacts import WorkflowCapabilityRef
 from wf_platform import CapabilityRef
 
 WorkflowSurfaceCapabilityId: TypeAlias = CapabilityRef | WorkflowCapabilityRef
 
+_CAPABILITY_REF_ADAPTER = TypeAdapter(CapabilityRef)
+_WORKFLOW_CAPABILITY_REF_ADAPTER = TypeAdapter(WorkflowCapabilityRef)
+
 
 def parse_workflow_surface_capability_id(
     value: str | dict[str, Any],
@@ -19,10 +24,10 @@ def parse_workflow_surface_capability_id(
     """
     if isinstance(value, dict):
         if "artifact_id" in value and "version" in value:
-            return WorkflowCapabilityRef._validate(value)
-        return CapabilityRef._validate(value)
+            return _WORKFLOW_CAPABILITY_REF_ADAPTER.validate_python(value)
+        return _CAPABILITY_REF_ADAPTER.validate_python(value)
 
     try:
         return WorkflowCapabilityRef.parse(value)
-    except ValueError:
+    except TypeError, ValueError, ValidationError:
         return CapabilityRef.parse(value)

src/wf_api/runs.py

@@ -57,6 +57,7 @@ class WorkflowRunApi:
         workflow_input: dict[str, Any],
         trace_range: TraceRangeLike | None = None,
     ) -> dict[str, Any]:
+        trace_values = _trace_range_values(trace_range)
         deployment, artifact, diagnostics, tree = (
             self.deployments.deployment_validation(deployment_id)
         )
@@ -96,22 +97,7 @@ class WorkflowRunApi:
             error=run.error,
             output=run.output,
             trace_count=len(run.trace),
-            trace=(
-                [
-                    asdict(entry)
-                    for entry in run.trace[
-                        trace_range.start : trace_range.start + trace_range.limit
-                    ]
-                ]
-                if trace_range is not None
-                else None
-            ),
-            trace_start=trace_range.start if trace_range is not None else None,
-            trace_limit=trace_range.limit if trace_range is not None else None,
-            trace_truncated=(
-                trace_range is not None
-                and len(run.trace) > trace_range.start + trace_range.limit
-            ),
+            **_trace_slice_fields(run, trace_values),
         )
 
     async def resume_run(
@@ -123,6 +109,7 @@ class WorkflowRunApi:
         trace_range: TraceRangeLike | None = None,
     ) -> dict[str, Any]:
         """Resume one durable interrupted deployment run."""
+        trace_values = _trace_range_values(trace_range)
         record, stopped_run = restore_interrupted_run(self._run_store(), run_id)
         environment = record.environment
         diagnostics = validate_pinned_resume_environment(
@@ -176,22 +163,7 @@ class WorkflowRunApi:
             error=run.error,
             output=run.output,
             trace_count=len(run.trace),
-            trace=(
-                [
-                    asdict(entry)
-                    for entry in run.trace[
-                        trace_range.start : trace_range.start + trace_range.limit
-                    ]
-                ]
-                if trace_range is not None
-                else None
-            ),
-            trace_start=trace_range.start if trace_range is not None else None,
-            trace_limit=trace_range.limit if trace_range is not None else None,
-            trace_truncated=(
-                trace_range is not None
-                and len(run.trace) > trace_range.start + trace_range.limit
-            ),
+            **_trace_slice_fields(run, trace_values),
         )
 
     async def inspect_run(self, *, run_id: str) -> dict[str, Any]:
@@ -219,9 +191,9 @@ class WorkflowRunApi:
         trace_range: TraceRangeLike,
     ) -> dict[str, Any]:
         """Return only a caller-bounded debug trace slice from a stopped run."""
+        trace_values = _trace_range_values(trace_range)
         record, run = load_stored_run(self._run_store(), run_id)
         environment = record.environment
-        end = trace_range.start + trace_range.limit
         return _run_payload(
             deployment=environment.deployment,
             artifact=environment.root_artifact,
@@ -230,13 +202,42 @@ class WorkflowRunApi:
             resume_readiness=record.resume_readiness.value,
             diagnostics=record.diagnostics,
             trace_count=len(run.trace),
-            trace=[asdict(entry) for entry in run.trace[trace_range.start : end]],
-            trace_start=trace_range.start,
-            trace_limit=trace_range.limit,
-            trace_truncated=len(run.trace) > end,
+            **_trace_slice_fields(run, trace_values),
         )
 
 
+def _trace_range_values(
+    trace_range: TraceRangeLike | None,
+) -> tuple[int, int] | None:
+    """Validate protocol-level trace ranges before they reach Python slicing."""
+    if trace_range is None:
+        return None
+    start = trace_range.start
+    limit = trace_range.limit
+    if start < 0:
+        raise ValueError("trace_range.start must be >= 0")
+    if limit <= 0:
+        raise ValueError("trace_range.limit must be > 0")
+    return start, limit
+
+
+def _trace_slice_fields(
+    run: RunState,
+    trace_range: tuple[int, int] | None,
+) -> dict[str, Any]:
+    """Return bounded trace payload fields, or no trace fields when omitted."""
+    if trace_range is None:
+        return {}
+    start, limit = trace_range
+    end = start + limit
+    return {
+        "trace": [asdict(entry) for entry in run.trace[start:end]],
+        "trace_start": start,
+        "trace_limit": limit,
+        "trace_truncated": len(run.trace) > end,
+    }
+
+
 def _run_payload(
     *,
     deployment: WorkflowDeployment,

src/wf_api/wrapper_hints.py

@@ -110,18 +110,16 @@ def wrapper_hints_for_capability(
         output_schema, output_properties
     )
     output_map = {name: f"state.{name}" for name in sorted(output_map_properties)}
-    state_schema = {
-        "type": "object",
-        "properties": {
-            name: schema for name, schema in sorted(output_map_properties.items())
-        },
-    }
-    wrapper_output_schema = {
+    mapped_output_schema = {
         "type": "object",
         "properties": {
             name: schema for name, schema in sorted(output_map_properties.items())
         },
     }
+    # The minimal wrapper stores mapped outputs in state and returns the same
+    # fields. Split this later only when wrappers support separate return shape.
+    state_schema = mapped_output_schema
+    wrapper_output_schema = mapped_output_schema
     missing_decisions = _missing_decisions_for_output(hint_output_schema)
     outcome_candidates = _boolean_outcome_candidates(output_properties)
     if outcome_candidates:

src/wf_mcp/broker/service/workflow_live_checks.py

@@ -30,7 +30,6 @@ from .core import WfMcpService
 
 LIVE_SOURCE_CHECK_TIMEOUT_SECONDS = 8.0
 _LIVE_SOURCE_CHECK_FAILURES = (
-    KeyError,
     TimeoutError,
     OSError,
     anyio.ClosedResourceError,
@@ -80,6 +79,16 @@ async def live_source_diagnostics(
             continue
         try:
             connection = service.connections.get(source_id)
+        except KeyError as exc:
+            diagnostics.append(
+                _source_unreachable_diagnostic(
+                    logical_ref=logical_ref,
+                    source_id=source_id,
+                    exc=exc,
+                )
+            )
+            continue
+        try:
             adapter = require_adapter(connection, service.adapters)
             auth = service.load_auth(source_id)
             await asyncio.wait_for(
@@ -88,7 +97,23 @@ async def live_source_diagnostics(
             )
         except _LIVE_SOURCE_CHECK_FAILURES as exc:
             diagnostics.append(
-                DependencyDiagnostic(
+                _source_unreachable_diagnostic(
+                    logical_ref=logical_ref,
+                    source_id=source_id,
+                    exc=exc,
+                )
+            )
+    return diagnostics
+
+
+def _source_unreachable_diagnostic(
+    *,
+    logical_ref: str,
+    source_id: str,
+    exc: BaseException,
+) -> DependencyDiagnostic:
+    """Build a liveness diagnostic without catching unrelated probe bugs."""
+    return DependencyDiagnostic(
         severity=DiagnosticSeverity.ERROR,
         code="source_unreachable",
         logical_ref=logical_ref,
@@ -102,8 +127,6 @@ async def live_source_diagnostics(
             "configuration, or bind this deployment to another source."
         ),
     )
-            )
-    return diagnostics
 
 
 __all__ = [

src/wf_mcp/workflow_surface/handlers.py

@@ -16,11 +16,11 @@ from wf_api.artifacts import WorkflowArtifactApi
 from wf_api.capabilities import WorkflowCapabilityApi
 from wf_api.deployments import WorkflowDeploymentApi
 from wf_api.drafts import WorkflowDraftApi
+from wf_api.listing import paged_list_payload
 from wf_api.models import RawWorkflowPlan
 from wf_api.runs import WorkflowRunApi
 
 from ..broker.service.workflow_operation_context import context_from_service
-from wf_api.listing import paged_list_payload
 from .models import TraceRange
 
 if TYPE_CHECKING:

tests/wf_api/test_capability_api.py

@@ -2,6 +2,8 @@ from __future__ import annotations
 
 import asyncio
 
+import pytest
+
 from wf_artifacts import FileWorkflowArtifactStore
 from wf_api.capabilities import WorkflowCapabilityApi
 from wf_mcp.broker import WfMcpService
@@ -91,11 +93,8 @@ def test_inspect_capability_raises_on_unknown() -> None:
     )
     api, _service = _capability_api(artifact_store, register_echo=True)
 
-    try:
+    with pytest.raises(KeyError, match="no.such.capability"):
         asyncio.run(api.inspect_capability(qualified_name="no.such.capability"))
-        assert False, "expected KeyError"
-    except KeyError as exc:
-        assert "no.such.capability" in str(exc)
 
 
 def test_call_capability_node_spec_success() -> None:

tests/wf_api/test_drafts_service.py

@@ -286,7 +286,7 @@ def test_validate_draft_workspace_refreshes_status() -> None:
     assert fetched["status"] == "invalid"
 
 
-def test_create_minimal_draft_workspace_with_error_route() -> None:
+def test_create_minimal_draft_workspace_minimal_success_path() -> None:
     artifact_store = FileWorkflowArtifactStore(
         local_temp_root() / "drafts_minimal_workspace"
     )

tests/wf_api/test_import_direction.py

@@ -15,15 +15,13 @@ def test_wf_api_has_no_wf_mcp_imports() -> None:
         tree = ast.parse(py_file.read_text(encoding="utf-8"), filename=str(py_file))
         for node in ast.walk(tree):
             if isinstance(node, ast.ImportFrom) and node.module is not None:
-                if node.module.startswith("wf_mcp") or node.module.startswith(
-                    "wf_mcp."
-                ):
+                if node.module == "wf_mcp" or node.module.startswith("wf_mcp."):
                     violations.append(
                         f"{module}:{node.lineno}: from {node.module} import ..."
                     )
             elif isinstance(node, ast.Import):
                 for alias in node.names:
-                    if alias.name.startswith("wf_mcp"):
+                    if alias.name == "wf_mcp" or alias.name.startswith("wf_mcp."):
                         violations.append(
                             f"{module}:{node.lineno}: import {alias.name}"
                         )

tests/wf_api/test_run_api.py

@@ -3,6 +3,8 @@ from __future__ import annotations
 import asyncio
 from pathlib import Path
 
+import pytest
+
 from wf_artifacts import FileWorkflowArtifactStore, WorkflowDeployment
 from wf_api.runs import WorkflowRunApi
 from wf_mcp.broker import WfMcpService
@@ -159,6 +161,29 @@ def test_run_api_inspect_and_bounded_trace() -> None:
     assert trace["trace_count"] == summary["trace_count"]
 
 
+def test_run_api_rejects_invalid_trace_range_before_store_lookup() -> None:
+    root = local_temp_root() / "run_api_invalid_trace_range"
+    service, _ = _service_with_echo(root)
+    context = context_from_service(service)
+    api = WorkflowRunApi(context)
+
+    with pytest.raises(ValueError, match="trace_range.start"):
+        asyncio.run(
+            api.read_run_trace(
+                run_id="missing",
+                trace_range=SimpleTraceRange(start=-1, limit=1),
+            )
+        )
+
+    with pytest.raises(ValueError, match="trace_range.limit"):
+        asyncio.run(
+            api.read_run_trace(
+                run_id="missing",
+                trace_range=SimpleTraceRange(start=0, limit=0),
+            )
+        )
+
+
 def test_run_api_handler_delegation_matches() -> None:
     root = local_temp_root() / "run_api_delegation"
     service, _ = _service_with_echo(root)

tests/wf_mcp/workflow_surface/test_next_actions.py

@@ -130,6 +130,18 @@ def test_next_actions_from_failed_run_recommends_bounded_trace() -> None:
     assert dumped["patch_examples"][0]["tool"] == NextActionTool.READ_RUN_TRACE.value
     assert dumped["patch_examples"][0]["request"]["run_id"] == "run_123"
     assert dumped["patch_examples"][0]["request"]["trace_range"]["start"] == 0
+    assert dumped["patch_examples"][0]["request"]["trace_range"]["limit"] == 12
+
+
+def test_next_actions_from_failed_run_caps_large_trace_example() -> None:
+    actions = NextActions.from_run_result(
+        run_id="run_123",
+        status="failed",
+        trace_count=100,
+        diagnostics=[],
+    )
+
+    dumped = actions.model_dump(mode="json")
     assert dumped["patch_examples"][0]["request"]["trace_range"]["limit"] == 25