fix(mcp): capture only a scalar projection of extra on $identify events

The v2 adapters hand the raw request ctx to identify/event_properties/
intent_fallback callbacks via `extra` so hosts can read headers — but
handle_identify embedded the whole dict into the captured $identify
parameters, where the sanitizer leaves opaque objects untouched and
truncation stringifies them: whatever the context repr carries (headers,
transport state) would ship to PostHog without key-based redaction.

Callbacks keep the full extra; captured parameters now carry only
JSON-safe scalars (e.g. session_id).

Generated-By: PostHog Code
Task-Id: ebafcb71-b03b-443d-b40c-d527ed4a04f4

Author: gesh

posthog/mcp/_internal.py

@@ -132,6 +132,24 @@ async def _maybe_await(value: Any) -> Any:
     return value
 
 
+def _captured_extra(extra: Optional[Dict[str, Any]]) -> Optional[Dict[str, Any]]:
+    """Project ``extra`` down to JSON-safe scalars before it is captured.
+
+    Callbacks (``identify``, ``event_properties``, ``intent_fallback``) receive
+    the full dict — on MCP SDK v2 that includes the raw request ``ctx`` so hosts
+    can read headers. Captured event parameters must not: the sanitizer leaves
+    opaque objects untouched and truncation stringifies them, which would ship
+    whatever the object's repr carries (headers, auth material, transport state)
+    to PostHog without key-based redaction."""
+    if extra is None:
+        return None
+    return {
+        key: value
+        for key, value in extra.items()
+        if value is None or isinstance(value, (str, int, float, bool))
+    }
+
+
 async def handle_identify(
     data: MCPAnalyticsData,
     session_id: str,
@@ -170,7 +188,7 @@ async def handle_identify(
                 "session_id": session_id,
                 "resource_name": _get_request_resource_name(request),
                 "event_type": MCPAnalyticsEventType.IDENTIFY,
-                "parameters": {"request": request, "extra": extra},
+                "parameters": {"request": request, "extra": _captured_extra(extra)},
                 "timestamp": datetime.now(timezone.utc),
             }
     except Exception as error:  # noqa: BLE001

posthog/test/mcp/test_v2_mcpserver.py

@@ -208,6 +208,37 @@ async def test_identify_sets_distinct_id_and_groups():
     assert _events(client, "$identify")
 
 
+async def test_identify_callback_sees_ctx_but_capture_does_not():
+    """The v2 adapters hand the raw request ``ctx`` to callbacks via ``extra`` so
+    hosts can read headers — but the captured $identify parameters must carry
+    only a scalar projection, or the stringified context (headers, transport
+    state) would ship to PostHog without key-based redaction."""
+    server = make_server()
+    client = FakeClient()
+    seen = {}
+
+    def identify(request, extra):
+        seen["extra"] = extra
+        return UserIdentity(distinct_id="user_9")
+
+    instrument(server, client, MCPAnalyticsOptions(identify=identify))
+
+    await _call_tool(
+        server, "add", {"a": 1, "b": 1, "context": "identity capture check"}
+    )
+    await _flush()
+
+    # the callback got the live context object...
+    assert seen["extra"]["ctx"] is not None
+
+    # ...but the captured event only carries JSON-safe scalars
+    identify_events = _events(client, "$identify")
+    assert identify_events
+    captured_extra = identify_events[0]["properties"]["$mcp_parameters"]["extra"]
+    assert "ctx" not in captured_extra
+    assert set(captured_extra) <= {"session_id"}
+
+
 async def test_report_missing_advertises_and_captures():
     server = make_server()
     client = FakeClient()