12 KiB
MCP Runtime Read Resource 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: Add persistent MCP runtime support for read_resource() by routing it through the existing owner-task operation queue and McpSourceClient.
Architecture: This is the first non-tool runtime operation. It must stay a thin wrapper over McpSourceClient.read_resource() so raw MCP SDK types and transport ownership do not leak. PersistentMcpSession and McpRuntimePool gain read_resource; get_prompt, invoke_method, and send_notification remain out of scope.
Tech Stack: Python 3.14, MCP Python SDK, pytest/pytest-asyncio, ruff, basedpyright.
File Structure
- Modify
src/wf_sources_mcp/runtime/factory.py- Add
_SessionOwner.read_resource(uri). - Implement through
submit(operation="read_resource", run=lambda client: client.read_resource(uri)).
- Add
- Modify
src/wf_sources_mcp/runtime/session.py- Add
RawResourceReadercallback type. - Add
read_resource_callbackfield. - Add
read_resource(uri)method. - Keep fallback
client.read_resource()serialization for injected/fake sessions.
- Add
- Modify
src/wf_sources_mcp/runtime/pool.py- Add
McpRuntimePool.read_resource(connection, auth, uri).
- Add
- Modify tests:
tests/wf_sources_mcp/test_runtime.pytests/wf_mcp/test_stateful_runtime.pyonly if type compatibility requires it.
- Update docs:
docs/current_roadmap.mddocs/superpowers/specs/2026-06-03-long-lived-workflow-api-boundary.md
Hard Boundaries
- Do not add persistent runtime
get_prompt,invoke_method, orsend_notification. - Do not expose raw
ClientSessionoutside the owner task. - Do not return raw MCP SDK result models from
read_resource; return serializeddict[str, Any]. - Do not alter
McpSourceClient.read_resource()behavior. - Do not dispatch by string;
operation="read_resource"is metadata only.
Task 1: Add Persistent Read Tests
Files:
-
Modify:
tests/wf_sources_mcp/test_runtime.py -
Step 1: Extend fake client with resource reads
In tests/wf_sources_mcp/test_runtime.py, import AnyUrl:
from pydantic import AnyUrl
Update _FakeFactory._create_with_stack()'s _FakeClient class:
class _FakeClient:
async def call_tool(
self, tool_name: str, payload: dict[str, object]
) -> RawCallToolResult:
return await factory._call_tool(tool_name, payload)
async def read_resource(self, uri: AnyUrl):
return type(
"ReadResourceResult",
(),
{
"model_dump": lambda _self, **_kwargs: {
"contents": [{"uri": str(uri), "text": "resource text"}]
}
},
)()
- Step 2: Add session-level persistent read test
Append to tests/wf_sources_mcp/test_runtime.py:
@pytest.mark.asyncio
async def test_persistent_session_factory_routes_resource_reads_through_owner() -> None:
factory = _FakeFactory()
connection = _connection()
session = await factory.create(connection, None)
await session.call_tool("echo", {"text": "one"})
resource_payload = await session.read_resource("fixture://docs/welcome")
await session.close()
assert factory.created_connections == [connection]
assert factory.calls == [("echo", {"text": "one"})]
assert resource_payload == {
"contents": [
{"uri": "fixture://docs/welcome", "text": "resource text"},
]
}
- Step 3: Add pool-level persistent read test
Append:
@pytest.mark.asyncio
async def test_runtime_pool_reuses_session_for_tool_and_resource_read() -> None:
factory = _FakeFactory()
pool = McpRuntimePool(factory.create)
connection = _connection()
tool_result = await pool.call_tool(connection, None, "echo", {"text": "one"})
resource_payload = await pool.read_resource(
connection,
None,
"fixture://docs/welcome",
)
await pool.close_all()
assert tool_result.output == {"echoed": "one"}
assert resource_payload["contents"][0]["text"] == "resource text"
assert factory.created_connections == [connection]
- Step 4: Update public surface test
Update test_persistent_session_public_runtime_is_tool_call_only to allow read_resource and still forbid the rest:
def test_persistent_session_public_runtime_exposes_only_tool_and_resource_read() -> None:
public_operations = {
name
for name in dir(PersistentMcpSession)
if not name.startswith("_") and callable(getattr(PersistentMcpSession, name))
}
assert "call_tool" in public_operations
assert "read_resource" in public_operations
assert "get_prompt" not in public_operations
assert "invoke_method" not in public_operations
assert "send_notification" not in public_operations
- Step 5: Run tests and confirm failure
Run:
uv run pytest tests/wf_sources_mcp/test_runtime.py -q
Expected: fail because PersistentMcpSession.read_resource and McpRuntimePool.read_resource do not exist yet.
Task 2: Add read_resource to Persistent Session
Files:
-
Modify:
src/wf_sources_mcp/runtime/session.py -
Step 1: Update imports and callback types
Add AnyUrl import:
from pydantic import AnyUrl
Add callback type:
RawResourceReader = Callable[[str], Awaitable[dict[str, Any]]]
- Step 2: Add callback field
In PersistentMcpSession, add:
read_resource_callback: RawResourceReader | None = None
- Step 3: Add
read_resourcemethod
Add below call_tool():
async def read_resource(self, uri: str) -> dict[str, Any]:
"""Read an MCP resource through the owner task or injected session."""
if self.read_resource_callback is not None:
return await self.read_resource_callback(uri)
if self.client is not None:
result = await self.client.read_resource(AnyUrl(uri))
return result.model_dump(by_alias=True, mode="json", exclude_none=True)
raise RuntimeError("persistent MCP session has no resource read transport")
- Step 4: Run source tests
Run:
uv run pytest tests/wf_sources_mcp/test_runtime.py -q
Expected: still fail at factory/pool routing because callbacks are not wired yet.
Task 3: Route read_resource Through Owner Queue
Files:
-
Modify:
src/wf_sources_mcp/runtime/factory.py -
Step 1: Wire callback in
PersistentSessionFactory.create()
Update returned PersistentMcpSession:
return PersistentMcpSession(
connection=connection,
auth=auth,
call_callback=owner.call_tool,
read_resource_callback=owner.read_resource,
close_callback=owner.close,
)
- Step 2: Add owner
read_resource()
Add below _SessionOwner.call_tool():
async def read_resource(self, uri: str) -> dict[str, Any]:
"""Submit a resource read through the generic owner-task operation queue."""
return await self.submit(
operation="read_resource",
run=lambda client: client.read_resource(uri),
)
- Step 3: Run source tests
Run:
uv run pytest tests/wf_sources_mcp/test_runtime.py -q
Expected: pool test still fails until pool method is added; direct session read should pass.
Task 4: Add read_resource to Runtime Pool
Files:
-
Modify:
src/wf_sources_mcp/runtime/pool.py -
Step 1: Add pool method
Add below call_tool():
async def read_resource(
self,
connection: McpSourceConnection,
auth: AuthRecord | None,
uri: str,
) -> dict[str, Any]:
session = await self.get_session(connection, auth)
return await session.read_resource(uri)
- Step 2: Run runtime tests
Run:
uv run pytest tests/wf_sources_mcp/test_runtime.py tests/wf_mcp/test_stateful_runtime.py -q
Expected: pass.
Task 5: Verify Surface and Update Docs
Files:
-
Modify:
docs/current_roadmap.md -
Modify:
docs/superpowers/specs/2026-06-03-long-lived-workflow-api-boundary.md -
Move after completion:
docs/superpowers/plans/2026-06-07-mcp-runtime-read-resource.mdtodocs/historical/superpowers/plans/2026-06-07-mcp-runtime-read-resource.md -
Step 1: Verify forbidden runtime methods remain absent
Run:
rg -n "def (list_tools|list_resources|list_prompts|get_prompt|invoke_method|send_notification)" src/wf_sources_mcp/runtime
Expected: no matches. read_resource is now allowed.
- Step 2: Update roadmap
In docs/current_roadmap.md, under the MCP upstream source runtime cleanup bullets, add:
- Completed: persistent MCP runtime can now route `read_resource` through
the owner-task queue and `McpSourceClient`. This is intentionally a thin
wrapper over the existing source-client facade; prompt/raw method runtime
operations remain separate future slices.
- Step 3: Update long-lived API boundary spec
In docs/superpowers/specs/2026-06-03-long-lived-workflow-api-boundary.md, add after the runtime operation queue item:
11. Complete: persistent MCP runtime can route `read_resource` through the
owner-task queue and `McpSourceClient`. Runtime still does not expose
`get_prompt`, raw method invocation, or notifications.
Renumber following item if needed.
- Step 4: Archive completed plan after implementation
Run only after all code/tests pass:
git mv docs/superpowers/plans/2026-06-07-mcp-runtime-read-resource.md docs/historical/superpowers/plans/2026-06-07-mcp-runtime-read-resource.md
Task 6: Final Verification
Files:
-
All changed files.
-
Step 1: Run focused tests
Run:
uv run pytest tests/wf_sources_mcp tests/wf_mcp/test_stateful_runtime.py tests/wf_mcp/test_compat_imports.py -q
Expected: pass.
- Step 2: Run source typecheck
Run:
uv run basedpyright --level error src
Expected: 0 errors, 0 warnings, 0 notes.
- Step 3: Run focused test typecheck
Run:
uv run basedpyright --level error tests/wf_sources_mcp/test_runtime.py tests/wf_mcp/test_stateful_runtime.py
Expected: 0 errors, 0 warnings, 0 notes.
- Step 4: Run focused lint
Run:
uv run ruff check src/wf_sources_mcp/runtime tests/wf_sources_mcp/test_runtime.py tests/wf_mcp/test_stateful_runtime.py
Expected: All checks passed!
- Step 5: Confirm no forbidden runtime expansion
Run:
rg -n "def (list_tools|list_resources|list_prompts|get_prompt|invoke_method|send_notification)" src/wf_sources_mcp/runtime
Expected: no matches.
- Step 6: Check whitespace
Run:
git diff --check
Expected: no whitespace errors. CRLF warnings are acceptable on Windows.
Self-Review
- Spec coverage: The plan adds persistent resource reads through the existing generic queue and source-client facade while keeping the surface intentionally narrow.
- Placeholder scan: No placeholder steps remain.
- Type consistency:
read_resourcereturnsdict[str, Any]everywhere. - Hard boundary: no prompt/raw-method/notification runtime methods are added.