nested state path

docs/core_state_mapping_and_merge.md

@@ -140,9 +140,8 @@ creates a reusable boundary for:
 
 ## State Declarations and Merge Rules
 
-The current implementation attaches merge behavior to declared top-level state
-fields. The target model should later allow nested declared state paths while
-keeping the internal representation flat:
+State merge behavior is attached to declared exact state paths while keeping the
+internal representation flat:
 
 ```python
 fields = {
@@ -226,10 +225,12 @@ current state merge behavior.
 
 ### Phase 2: Nested declared state paths
 
-- allow exact nested state path declarations
-- resolve merge strategies by exact destination path only
-- keep undeclared paths as `replace`
-- keep `merge_object` shallow
+Implemented in core:
+
+- exact nested state path declarations
+- merge strategies resolved by exact destination path only
+- undeclared paths remain `replace`
+- `merge_object` remains shallow
 
 ### Phase 3: Reducer capabilities
 

docs/schema_validation.md

@@ -22,13 +22,13 @@ It still does not solve:
 - semantic compatibility between Pydantic-generated schemas and every possible
   external JSON Schema dialect
 - typed Python object creation from arbitrary JSON Schema
-- workflow state merge behavior
+- workflow state merge behavior beyond declared exact-path metadata
 - deep node-local map-path validation beyond statically knowable schema roots
 - better domain-specific error payloads beyond `WorkflowExecutionError`
 
-This means schema fields are mostly contracts for authoring, planning,
-documentation, and mapping validation today. They are not yet strong runtime
-guards.
+This means schema fields are contracts for authoring, planning, documentation,
+and mapping validation. Runtime state merge behavior is separate metadata on
+declared exact state paths; undeclared paths still use `replace`.
 
 ## Why This Matters
 

docs/superpowers/plans/2026-05-17-nested-state-paths.md

@@ -0,0 +1,61 @@
+# Nested State Paths Implementation Plan
+
+> **For agentic workers:** REQUIRED SUB-SKILL: Use superpowers:subagent-driven-development (recommended) or superpowers:executing-plans to implement this plan task-by-task. Steps use checkbox (`- [ ]`) syntax for tracking.
+
+**Goal:** Let `wf_core` declare nested workflow state paths and apply merge behavior by exact destination path.
+
+**Architecture:** Keep state declarations internally flat and path-keyed. Continue allowing undeclared state paths, but only exact declared paths receive typed merge behavior; parent declarations do not implicitly govern descendants. Reuse the state patch boundary added in Phase 1 and change only schema wording, validation coverage, runtime lookup, and docs.
+
+**Tech Stack:** Python, Pydantic, pytest, existing `wf_core` state runtime.
+
+---
+
+## File Structure
+
+- Modify `src/wf_core/models/schemas.py`
+  - clarify that `StateField` / `StateSchema` are path-keyed, not root-only
+- Modify `src/wf_core/runtime/ops/state.py`
+  - resolve merge metadata by exact written state path
+- Add `tests/core/test_nested_state_paths.py`
+  - exact nested path merge behavior
+  - ancestor declarations do not govern descendants
+  - undeclared nested paths still replace
+- Update `docs/core_state_mapping_and_merge.md`
+  - mark nested declared state paths as implemented
+- Update `docs/schema_validation.md`
+  - clarify that runtime state merge metadata is now exact-path capable
+
+### Task 1: Pin Exact-Path State Behavior
+
+- [ ] Add failing tests proving:
+  - `state.person.tags` uses a declaration for `"person.tags"` with `append`
+  - a declaration for `"person"` does not cause `state.person.tags` to inherit `merge_object`
+  - undeclared `state.person.tags` defaults to `replace`
+- [ ] Run the focused tests and confirm the nested exact-path cases fail under the current root-only lookup.
+
+### Task 2: Implement Exact-Path Lookup
+
+- [ ] Update state model docstrings to describe path-keyed declarations.
+- [ ] In `write_state_value()`, look up `workflow.state_schema.fields[".".join(parts)]` instead of only the first path segment.
+- [ ] Keep undeclared paths as `replace`.
+- [ ] Run the focused tests and confirm they pass.
+
+### Task 3: Keep Docs Honest
+
+- [ ] Update the core mapping design doc so Phase 2 is recorded as implemented, not future work.
+- [ ] Update schema-validation docs to note exact-path state metadata without broadening payload validation claims.
+
+### Task 4: Verify
+
+- [ ] Run `uv run --with pytest pytest tests/core -q`
+- [ ] Run `uv run --with pytest pytest -q`
+- [ ] Run `uv run basedpyright --level error`
+- [ ] Call out any residual type-check failures that are unrelated to this work.
+
+## Non-Goals
+
+- flatten nested authoring schemas from `wf_authoring`
+- reducer registries or custom reducer capabilities
+- changing `merge_object` from shallow to deep
+- making parent declarations inherit into child paths
+- `START` token model changes

src/wf_authoring/ops/values.py

@@ -141,7 +141,11 @@ def pick_path(input: PickPathInput) -> ValueOutput:
 def project_fields(input: ProjectFieldsInput) -> MappingOutput:
     """Return only the requested existing fields from a mapping."""
     return MappingOutput(
-        mapping={field: input.mapping[field] for field in input.fields if field in input.mapping}
+        mapping={
+            field: input.mapping[field]
+            for field in input.fields
+            if field in input.mapping
+        }
     )
 
 

src/wf_core/models/schemas.py

@@ -17,7 +17,7 @@ class SchemaRef(BaseModel):
 
 
 class StateField(BaseModel):
-    """Declared root state field plus its runtime merge behavior."""
+    """Declared state path plus its runtime merge behavior."""
 
     type: str
     merge_strategy: Literal["replace", "append", "merge_object"] = "replace"
@@ -26,7 +26,7 @@ class StateField(BaseModel):
 
 
 class StateSchema(BaseModel):
-    """Workflow state schema keyed by declared root field name."""
+    """Workflow state schema keyed by declared exact state path."""
 
     model_config = ConfigDict(extra="allow")
 

src/wf_core/runtime/ops/merges.py

@@ -23,15 +23,23 @@ def apply_builtin_merge(
 
     if strategy == "append":
         if current_value is None:
-            return [incoming_value] if not isinstance(incoming_value, list) else incoming_value
+            return (
+                [incoming_value]
+                if not isinstance(incoming_value, list)
+                else incoming_value
+            )
         if not isinstance(current_value, list):
             raise WorkflowExecutionError(
                 f"cannot append into non-list state path {destination_path!r}"
             )
-        return [
-            *current_value,
-            *incoming_value,
-        ] if isinstance(incoming_value, list) else [*current_value, incoming_value]
+        return (
+            [
+                *current_value,
+                *incoming_value,
+            ]
+            if isinstance(incoming_value, list)
+            else [*current_value, incoming_value]
+        )
 
     if strategy == "merge_object":
         if current_value is None:

src/wf_core/runtime/ops/state.py

@@ -39,7 +39,9 @@ def apply_mapped_state(
     missing_field_message: str,
 ) -> dict[str, Any]:
     if has_overlapping_paths(mapping.values()):
-        raise WorkflowExecutionError("mapped state patch has overlapping destination paths")
+        raise WorkflowExecutionError(
+            "mapped state patch has overlapping destination paths"
+        )
 
     patch: dict[str, Any] = {}
     for source_field, destination_path in mapping.items():
@@ -69,8 +71,8 @@ def write_state_value(
             f"executor only supports writes into state.*, got {destination_path!r}"
         )
 
-    field_name = parts[0]
-    declared_field = workflow.state_schema.fields.get(field_name)
+    declared_path = ".".join(parts)
+    declared_field = workflow.state_schema.fields.get(declared_path)
     merge_strategy = declared_field.merge_strategy if declared_field else "replace"
     key_path = parts
     current_value = get_nested_value(state, key_path)

tests/authoring/test_ops.py

@@ -268,7 +268,10 @@ def test_project_fields_selects_named_keys() -> None:
     registry = build_registry(project_fields)
 
     result = registry["authoring.project_fields"](
-        {"mapping": {"status": "done", "message": "ok", "debug": True}, "fields": ["status", "message"]},
+        {
+            "mapping": {"status": "done", "message": "ok", "debug": True},
+            "fields": ["status", "message"],
+        },
         RuntimeContext(current_node_id="project_fields"),
     )
 
@@ -282,7 +285,10 @@ def test_rename_fields_remaps_existing_keys() -> None:
     registry = build_registry(rename_fields)
 
     result = registry["authoring.rename_fields"](
-        {"mapping": {"provider_status": "done", "provider_message": "ok"}, "renames": {"provider_status": "status", "provider_message": "message"}},
+        {
+            "mapping": {"provider_status": "done", "provider_message": "ok"},
+            "renames": {"provider_status": "status", "provider_message": "message"},
+        },
         RuntimeContext(current_node_id="rename_fields"),
     )
 

tests/core/test_mapping_validation.py

@@ -21,7 +21,9 @@ def test_validation_rejects_overlapping_node_input_destinations() -> None:
         out_map={},
     ).validate_structure()
 
-    assert any("overlapping node-local input paths" in issue.message for issue in report.errors)
+    assert any(
+        "overlapping node-local input paths" in issue.message for issue in report.errors
+    )
 
 
 def test_validation_rejects_overlapping_state_write_destinations() -> None:
@@ -33,7 +35,10 @@ def test_validation_rejects_overlapping_state_write_destinations() -> None:
         },
     ).validate_structure()
 
-    assert any("overlapping state destination paths" in issue.message for issue in report.errors)
+    assert any(
+        "overlapping state destination paths" in issue.message
+        for issue in report.errors
+    )
 
 
 def _workflow(*, in_map: dict[str, str], out_map: dict[str, str]) -> Workflow:

tests/core/test_nested_state_paths.py

@@ -0,0 +1,48 @@
+from __future__ import annotations
+
+from wf_core import SchemaRef, StateField, StateSchema, Workflow
+from wf_core.runtime.ops.state import write_state_value
+
+
+def test_exact_nested_state_path_uses_declared_merge_strategy() -> None:
+    workflow = _workflow(
+        fields={"person.tags": StateField(type="array", merge_strategy="append")}
+    )
+    state = {"person": {"tags": ["seed"]}}
+
+    write_state_value(workflow, state, "state.person.tags", ["next"])
+
+    assert state["person"]["tags"] == ["seed", "next"]
+
+
+def test_parent_state_declaration_does_not_apply_to_nested_write() -> None:
+    workflow = _workflow(
+        fields={"person": StateField(type="object", merge_strategy="merge_object")}
+    )
+    state = {"person": {"tags": ["seed"]}}
+
+    write_state_value(workflow, state, "state.person.tags", ["next"])
+
+    assert state["person"]["tags"] == ["next"]
+
+
+def test_undeclared_nested_state_path_defaults_to_replace() -> None:
+    workflow = _workflow(fields={})
+    state = {"person": {"tags": ["seed"]}}
+
+    write_state_value(workflow, state, "state.person.tags", ["next"])
+
+    assert state["person"]["tags"] == ["next"]
+
+
+def _workflow(*, fields: dict[str, StateField]) -> Workflow:
+    return Workflow(
+        name="nested_state_paths",
+        input_schema=SchemaRef(type="object", properties={}),
+        state_schema=StateSchema(fields=fields),
+        output_schema=SchemaRef(type="object", properties={}),
+        node_defs=[],
+        start="unused",
+        nodes=[],
+        edges=[],
+    )