Keep the native project catalogue bounded (JVNAUTOSCI-2737)

docs/engineering/workflow_first_jira_to_von_migration_jvnautosci_1223.md

@@ -98,6 +98,8 @@ to their retained original without needing Jira. Native MCP tools expose the
 same canonical services, including `task_list_projects`, `task_get_project`,
 `task_get_collection`, `task_get`, `task_search` and `task_get_source_archive`.
 `task_get` accepts legacy Jira keys as well as Von task concept IDs.
+Project listings return navigation metadata and descriptions; retrieve source
+archives, retained-document references and their histories with `task_get_project`.
 
 A local stdio connection can bind the trusted operator once in its server
 environment:

src/backend/services/task_project_service.py

@@ -23,6 +23,20 @@
 
 PROJECT_RECORD = "task_project"
 COLLECTION_RECORD = "task_collection"
+_PROJECT_LIST_FIELDS = (
+    "name",
+    "key",
+    "description",
+    "source_system",
+    "source_site",
+    "source_id",
+    "source_url",
+    "organisation_concept_id",
+    "domain_project_concept_id",
+    "collection_concept_ids",
+    "writer",
+    "cutover_state",
+)
 
 
 def _find_concept(concept_id):
@@ -60,10 +74,22 @@ def list_project_collections(project):
 
 
 def list_task_projects(*, limit=100, offset=0):
+    """List navigation metadata; archive details remain in get_task_project.
+
+    Retained source histories can grow independently of project navigation.
+    Do not load those histories and document manifests into every catalogue
+    response, where they can exceed the MCP transport's response size limit.
+    """
     query = {f"attributes.{PROJECT_RECORD}": {"$exists": True}}
     docs = ConceptsRepository.find(
         query,
-        projection={"concept_id": 1, "attributes.task_project": 1},
+        projection={
+            "concept_id": 1,
+            **{
+                f"attributes.{PROJECT_RECORD}.{field}": 1
+                for field in _PROJECT_LIST_FIELDS
+            },
+        },
         limit=max(1, min(int(limit), 200)),
         skip=max(0, int(offset)),
     )

tests/backend/test_task_project_service.py

@@ -1,10 +1,66 @@
 import copy
+import json
 
 import pytest
 
 from src.backend.services import task_project_service as projects
 
 
+def test_project_catalogue_does_not_grow_with_retained_source_history(monkeypatch):
+    record = {
+        "name": "Research project",
+        "key": "RESEARCH",
+        "description": "The complete project description",
+        "writer": "jira",
+        "source_id": "123",
+        "collection_concept_ids": ["#V#research_tasks"],
+        "source_archive_history": [{"manifest": "x" * 100_000}],
+        "external_document_archives": [{"manifest": "y" * 100_000}],
+    }
+    query_seen = {}
+
+    def find(query, *, projection, limit, skip):
+        query_seen.update(query=query, limit=limit, skip=skip)
+        selected = {
+            key: value
+            for key, value in record.items()
+            if projection.get("attributes.task_project")
+            or projection.get(f"attributes.task_project.{key}")
+        }
+        return [
+            {
+                "concept_id": f"#V#project_{index}",
+                "attributes": {"task_project": selected},
+            }
+            for index in range(skip, min(skip + limit, 43))
+        ]
+
+    monkeypatch.setattr(projects.ConceptsRepository, "find", find)
+    result = projects.list_task_projects()
+    assert len(result) == 43
+    assert len(json.dumps({"success": True, "projects": result})) < 100_000
+    assert result[0]["description"] == record["description"]
+    assert result[0]["collection_concept_ids"] == record["collection_concept_ids"]
+    assert result[0]["writer"] == "jira"
+    assert result[0]["source_id"] == "123"
+    assert projects.list_task_projects(limit=2, offset=41) == result[41:]
+    assert query_seen == {
+        "query": {"attributes.task_project": {"$exists": True}},
+        "limit": 2,
+        "skip": 41,
+    }
+
+    monkeypatch.setattr(projects, "can_access_concept", lambda key: True)
+    monkeypatch.setattr(
+        projects,
+        "get_concept_by_concept_id_exact",
+        lambda key: {"attributes": {"task_project": copy.deepcopy(record)}},
+    )
+    detail = projects.get_task_project("#V#project_0")
+    assert detail["source_archive_history"] == record["source_archive_history"]
+    assert detail["external_document_archives"] == record["external_document_archives"]
+
+
 def test_project_collection_rerun_preserves_identity_and_writer(monkeypatch):
     docs = {}
     publish_during_refresh = False