diff --git a/tests/integration/admin/test_e2e_admin_flow.py b/tests/integration/admin/test_e2e_admin_flow.py
new file mode 100644
index 0000000..28753d0
--- /dev/null
+++ b/tests/integration/admin/test_e2e_admin_flow.py
@@ -0,0 +1,560 @@
+"""End-to-end admin workflow integration tests (U10).
+
+Verifies complete admin workflows end-to-end via the FastAPI TestClient:
+
+- Department lifecycle: create → update → bind skill → bind KB →
+  disable → enable → unbind skill → delete.
+- User lifecycle: create → assign department → reset password →
+  remove department → soft delete.
+- LLM config lifecycle: add provider → set API key → set fallback →
+  set quota → list → delete provider.
+- Skill management flow: disable skill → verify excluded from
+  ``GET /skills`` → enable → verify included.
+- Usage dashboard flow: query usage endpoints with empty data →
+  verify empty results + CSV header.
+
+The tests mount a minimal FastAPI app with the ``admin_router`` (and
+the public ``skills`` router for the skill-management flow). The
+``_require_admin`` dependency is overridden so the tests don't need
+real JWTs.
+"""
+
+from __future__ import annotations
+
+import csv
+import io
+import os
+import uuid
+from pathlib import Path
+from typing import Any
+
+import pytest
+import yaml
+from fastapi import FastAPI, HTTPException
+from fastapi.testclient import TestClient
+
+from agentkit.server.admin.context import DepartmentContext
+from agentkit.server.admin.kb_service import set_kb_service
+from agentkit.server.admin.llm_config_service import (
+    LlmConfigService,
+    set_llm_config_service,
+)
+from agentkit.server.admin.skill_service import set_skill_service
+from agentkit.server.admin.usage_service import set_usage_service
+from agentkit.server.auth.models import init_auth_db
+from agentkit.server.auth.session_service import SessionService, set_session_service
+from agentkit.server.routes import admin as admin_routes_module
+from agentkit.server.routes import skills as skills_routes_module
+from agentkit.skills.registry import SkillRegistry
+
+
+# ---------------------------------------------------------------------------
+# Test data
+# ---------------------------------------------------------------------------
+
+
+_VALID_SKILL_YAML = """\
+name: e2e_test_skill
+agent_type: simple_generation
+version: "1.0.0"
+description: "E2E test skill"
+task_mode: llm_generate
+execution_mode: direct
+max_steps: 1
+prompt:
+  identity: "E2E"
+  instructions: "Handle test"
+tools: []
+"""
+
+
+def _sample_agentkit_config() -> dict[str, Any]:
+    """A minimal agentkit.yaml-style config for testing."""
+    return {
+        "server": {"host": "0.0.0.0", "port": 8001},
+        "llm": {
+            "providers": {
+                "openai": {
+                    "type": "openai",
+                    "api_key": "sk-test-12345678",
+                    "base_url": "https://api.openai.com/v1",
+                    "models": {"gpt-4o": {}},
+                    "max_tokens": 4096,
+                    "timeout": 120.0,
+                },
+            },
+            "model_aliases": {"gpt4": "openai/gpt-4o"},
+            "fallbacks": {},
+        },
+    }
+
+
+# ---------------------------------------------------------------------------
+# Fixtures
+# ---------------------------------------------------------------------------
+
+
+@pytest.fixture
+async def tmp_auth_db(tmp_path: Path) -> Path:
+    db_path = tmp_path / "e2e_admin.db"
+    await init_auth_db(db_path)
+    return db_path
+
+
+@pytest.fixture
+def tmp_agentkit_yaml(tmp_path: Path) -> Path:
+    """Create a temporary agentkit.yaml config file."""
+    path = tmp_path / "agentkit.yaml"
+    with open(path, "w", encoding="utf-8") as f:
+        yaml.dump(_sample_agentkit_config(), f, default_flow_style=False, allow_unicode=True)
+    return path
+
+
+@pytest.fixture
+def tmp_skills_dir(tmp_path: Path) -> str:
+    """A temp skills directory for YAML files."""
+    d = tmp_path / "skills"
+    d.mkdir()
+    return str(d)
+
+
+@pytest.fixture
+def skill_registry() -> SkillRegistry:
+    return SkillRegistry()
+
+
+@pytest.fixture
+def session_service(tmp_auth_db: Path):
+    """Install a SessionService singleton backed by the temp DB.
+
+    Required so that ``UserService.reset_password`` can find the
+    SessionService via ``get_session_service()`` and revoke sessions.
+    """
+    svc = SessionService(db_path=tmp_auth_db)
+    set_session_service(svc)
+    yield svc
+    set_session_service(None)
+
+
+@pytest.fixture(autouse=True)
+def _reset_singletons():
+    """Reset singletons before/after each test to avoid state leakage."""
+    set_llm_config_service(None)
+    set_skill_service(None)
+    set_kb_service(None)
+    set_usage_service(None)
+    yield
+    set_llm_config_service(None)
+    set_skill_service(None)
+    set_kb_service(None)
+    set_usage_service(None)
+
+
+def _make_admin_user() -> dict[str, Any]:
+    return {"user_id": "admin-1", "username": "admin", "role": "admin"}
+
+
+def _raise_forbidden() -> dict[str, Any]:
+    raise HTTPException(status_code=403, detail="Admin permission required")
+
+
+@pytest.fixture
+def admin_app(
+    tmp_auth_db: Path,
+    tmp_agentkit_yaml: Path,
+    tmp_skills_dir: str,
+    skill_registry: SkillRegistry,
+) -> FastAPI:
+    """A FastAPI app with admin + skills routers mounted.
+
+    The ``_require_admin`` dependency is overridden to return a fake
+    admin user. The :class:`LlmConfigService` singleton is pre-populated
+    so the routes use the temp YAML file.
+    """
+    set_llm_config_service(LlmConfigService(tmp_agentkit_yaml))
+
+    app = FastAPI()
+    app.state.auth_db_path = str(tmp_auth_db)
+    app.state.skill_registry = skill_registry
+
+    class _FakeServerConfig:
+        skill_paths = [tmp_skills_dir]
+
+    app.state.server_config = _FakeServerConfig()
+
+    app.include_router(admin_routes_module.admin_router, prefix="/api/v1")
+    app.include_router(skills_routes_module.router, prefix="/api/v1")
+
+    # Default: allow admin access.
+    app.dependency_overrides[admin_routes_module._require_admin] = lambda: _make_admin_user()
+    # Admin context for skills route (bypass department filtering).
+    app.dependency_overrides[skills_routes_module.get_department_context] = lambda: (
+        DepartmentContext(user_id="admin-1", department_ids=[], is_admin=True)
+    )
+    return app
+
+
+@pytest.fixture
+def admin_client(
+    admin_app: FastAPI,
+    session_service: SessionService,
+) -> TestClient:
+    """TestClient with admin access and SessionService installed."""
+    return TestClient(admin_app)
+
+
+# ---------------------------------------------------------------------------
+# Helper functions
+# ---------------------------------------------------------------------------
+
+
+def _create_department(client: TestClient, name: str, description: str = "") -> dict:
+    resp = client.post(
+        "/api/v1/admin/departments",
+        json={"name": name, "description": description},
+    )
+    assert resp.status_code == 201, resp.text
+    return resp.json()
+
+
+def _create_user(
+    client: TestClient,
+    *,
+    username: str,
+    email: str,
+    password: str = "Secret123!",
+    role: str = "member",
+    department_ids: list[str] | None = None,
+) -> dict:
+    payload: dict[str, Any] = {
+        "username": username,
+        "email": email,
+        "password": password,
+        "role": role,
+    }
+    if department_ids is not None:
+        payload["department_ids"] = department_ids
+    resp = client.post("/api/v1/admin/users", json=payload)
+    assert resp.status_code == 201, resp.text
+    return resp.json()
+
+
+def _write_skill_yaml(skills_dir: str, name: str, content: str) -> str:
+    path = os.path.join(skills_dir, f"{name}.yaml")
+    with open(path, "w", encoding="utf-8") as f:
+        f.write(content)
+    return path
+
+
+# ---------------------------------------------------------------------------
+# E2E admin flow tests
+# ---------------------------------------------------------------------------
+
+
+class TestE2EAdminFlow:
+    """End-to-end admin workflow test."""
+
+    def test_full_department_lifecycle(
+        self,
+        admin_client: TestClient,
+        tmp_skills_dir: str,
+        skill_registry: SkillRegistry,
+    ):
+        """Create department → update → bind skill → bind KB → disable →
+        enable → unbind skill → delete."""
+        # 1. Create department "HR".
+        hr = _create_department(admin_client, "HR", "Human Resources")
+        dept_id = hr["id"]
+        assert hr["name"] == "HR"
+        assert hr["is_active"] is True
+
+        # 2. GET /admin/departments → HR in list.
+        resp = admin_client.get("/api/v1/admin/departments")
+        assert resp.status_code == 200
+        names = {d["name"] for d in resp.json()}
+        assert "HR" in names
+
+        # 3. PATCH /admin/departments/{id} → update name to "Human Resources".
+        resp = admin_client.patch(
+            f"/api/v1/admin/departments/{dept_id}",
+            json={"name": "Human Resources", "description": "HR department"},
+        )
+        assert resp.status_code == 200
+        assert resp.json()["name"] == "Human Resources"
+
+        # 4. POST /admin/departments/{id}/skills/code_reviewer → bind skill.
+        resp = admin_client.post(f"/api/v1/admin/departments/{dept_id}/skills/code_reviewer")
+        assert resp.status_code == 201
+        assert resp.json()["skill_name"] == "code_reviewer"
+
+        # 5. GET /admin/departments/{id}/skills → ["code_reviewer"].
+        resp = admin_client.get(f"/api/v1/admin/departments/{dept_id}/skills")
+        assert resp.status_code == 200
+        assert resp.json() == ["code_reviewer"]
+
+        # 6. POST /admin/departments/{id}/kb/source-1 → bind KB.
+        resp = admin_client.post(f"/api/v1/admin/departments/{dept_id}/kb/source-1")
+        assert resp.status_code == 201
+        assert resp.json()["kb_source_id"] == "source-1"
+
+        # 7. GET /admin/departments/{id}/kb → ["source-1"].
+        resp = admin_client.get(f"/api/v1/admin/departments/{dept_id}/kb")
+        assert resp.status_code == 200
+        assert resp.json() == ["source-1"]
+
+        # 8. POST /admin/departments/{id}/disable → disabled.
+        resp = admin_client.post(f"/api/v1/admin/departments/{dept_id}/disable")
+        assert resp.status_code == 200
+        assert resp.json()["is_active"] is False
+
+        # 9. POST /admin/departments/{id}/enable → enabled.
+        resp = admin_client.post(f"/api/v1/admin/departments/{dept_id}/enable")
+        assert resp.status_code == 200
+        assert resp.json()["is_active"] is True
+
+        # 10. DELETE /admin/departments/{id}/skills/code_reviewer → unbound.
+        resp = admin_client.delete(f"/api/v1/admin/departments/{dept_id}/skills/code_reviewer")
+        assert resp.status_code == 200
+        assert resp.json() == {"unbound": True}
+        assert admin_client.get(f"/api/v1/admin/departments/{dept_id}/skills").json() == []
+
+        # 11. DELETE /admin/departments/{id} → deleted.
+        resp = admin_client.delete(f"/api/v1/admin/departments/{dept_id}")
+        assert resp.status_code == 200
+        assert resp.json() == {"deleted": True}
+        assert admin_client.get(f"/api/v1/admin/departments/{dept_id}").status_code == 404
+
+    def test_full_user_lifecycle(self, admin_client: TestClient):
+        """Create user → assign department → reset password → remove
+        department → delete."""
+        # 1. Create department "Engineering".
+        eng = _create_department(admin_client, "Engineering")
+        eng_id = eng["id"]
+
+        # 2. POST /admin/users → create user "alice".
+        alice = _create_user(
+            admin_client,
+            username="alice",
+            email="alice@example.com",
+        )
+        alice_id = alice["id"]
+        assert alice["username"] == "alice"
+        assert alice["departments"] == []
+
+        # 3. POST /admin/users/{id}/departments/{dept_id} → assign.
+        resp = admin_client.post(f"/api/v1/admin/users/{alice_id}/departments/{eng_id}")
+        assert resp.status_code == 201
+        assert resp.json() == {"assigned": True}
+
+        # 4. GET /admin/users/{id} → has department.
+        resp = admin_client.get(f"/api/v1/admin/users/{alice_id}")
+        assert resp.status_code == 200
+        body = resp.json()
+        assert len(body["departments"]) == 1
+        assert body["departments"][0]["name"] == "Engineering"
+
+        # 5. POST /admin/users/{id}/reset-password → reset.
+        resp = admin_client.post(
+            f"/api/v1/admin/users/{alice_id}/reset-password",
+            json={"new_password": "NewSecret456!"},
+        )
+        assert resp.status_code == 200
+        assert resp.json() == {"reset": True}
+
+        # 6. DELETE /admin/users/{id}/departments/{dept_id} → remove.
+        resp = admin_client.delete(f"/api/v1/admin/users/{alice_id}/departments/{eng_id}")
+        assert resp.status_code == 200
+        assert resp.json() == {"removed": True}
+
+        # Verify removal.
+        resp = admin_client.get(f"/api/v1/admin/users/{alice_id}")
+        assert resp.status_code == 200
+        assert resp.json()["departments"] == []
+
+        # 7. DELETE /admin/users/{id} → soft delete.
+        resp = admin_client.delete(f"/api/v1/admin/users/{alice_id}")
+        assert resp.status_code == 200
+        assert resp.json() == {"deleted": True}
+
+        # Second delete on the now-inactive user returns 404.
+        resp = admin_client.delete(f"/api/v1/admin/users/{alice_id}")
+        assert resp.status_code == 404
+
+    def test_llm_config_lifecycle(
+        self,
+        admin_client: TestClient,
+        tmp_agentkit_yaml: Path,
+    ):
+        """Add provider → set API key → set fallback → set quota →
+        delete provider."""
+        # 1. POST /admin/llm/providers → create "test-provider".
+        resp = admin_client.post(
+            "/api/v1/admin/llm/providers",
+            json={
+                "name": "test-provider",
+                "type": "openai",
+                "api_key": "sk-test-abcdef1234",
+                "base_url": "https://api.test.com",
+                "models": {"test-model": {}},
+                "max_tokens": 2048,
+                "timeout": 60.0,
+            },
+        )
+        assert resp.status_code == 201, resp.text
+        assert resp.json()["name"] == "test-provider"
+        assert resp.json()["api_key"].startswith("****")
+
+        # 2. POST /admin/llm/providers/test-provider/api-key → set key.
+        resp = admin_client.post(
+            "/api/v1/admin/llm/providers/test-provider/api-key",
+            json={"api_key": "sk-brand-new-key-9876"},
+        )
+        assert resp.status_code == 200
+        assert resp.json()["api_key"].startswith("****")
+
+        # 3. PUT /admin/llm/fallbacks/gpt-4 → set fallback chain.
+        resp = admin_client.put(
+            "/api/v1/admin/llm/fallbacks/gpt-4",
+            json={"chain": ["openai", "test-provider"]},
+        )
+        assert resp.status_code == 200
+        body = resp.json()
+        assert body["model"] == "gpt-4"
+        assert body["chain"] == ["openai", "test-provider"]
+
+        # 4. PUT /admin/departments/{id}/quotas → set token_limit.
+        dept_id = str(uuid.uuid4())
+        resp = admin_client.put(
+            f"/api/v1/admin/departments/{dept_id}/quotas",
+            json={
+                "quota_type": "token_limit",
+                "limit_value": 1000,
+                "period": "daily",
+            },
+        )
+        assert resp.status_code == 200
+        body = resp.json()
+        assert body["quota_type"] == "token_limit"
+        assert body["limit_value"] == 1000
+
+        # 5. GET /admin/llm/providers → provider in list.
+        resp = admin_client.get("/api/v1/admin/llm/providers")
+        assert resp.status_code == 200
+        names = {p["name"] for p in resp.json()}
+        assert "test-provider" in names
+        assert "openai" in names
+
+        # 6. DELETE /admin/llm/providers/test-provider → fails (used in fallback).
+        resp = admin_client.delete("/api/v1/admin/llm/providers/test-provider")
+        assert resp.status_code == 400
+        assert "fallback" in resp.json()["detail"].lower()
+
+        # Remove the fallback chain first, then delete.
+        admin_client.delete("/api/v1/admin/llm/fallbacks/gpt-4")
+        resp = admin_client.delete("/api/v1/admin/llm/providers/test-provider")
+        assert resp.status_code == 200
+        assert resp.json() == {"deleted": True}
+
+        # Confirm it's gone.
+        resp = admin_client.get("/api/v1/admin/llm/providers")
+        names = {p["name"] for p in resp.json()}
+        assert "test-provider" not in names
+
+    def test_skill_management_flow(
+        self,
+        admin_client: TestClient,
+        tmp_skills_dir: str,
+        skill_registry: SkillRegistry,
+    ):
+        """Disable skill → verify excluded → enable → verify included."""
+        # 1. Create a test skill YAML in tmp_skills_dir and register it.
+        _write_skill_yaml(tmp_skills_dir, "e2e_test_skill", _VALID_SKILL_YAML)
+        from agentkit.skills.loader import SkillLoader
+
+        SkillLoader(skill_registry=skill_registry).load_from_file(
+            os.path.join(tmp_skills_dir, "e2e_test_skill.yaml")
+        )
+
+        # Verify the skill is initially listed.
+        resp = admin_client.get("/api/v1/skills")
+        assert resp.status_code == 200
+        names = [s["name"] for s in resp.json()]
+        assert "e2e_test_skill" in names
+
+        # 2. POST /admin/skills/{name}/disable → disabled.
+        resp = admin_client.post("/api/v1/admin/skills/e2e_test_skill/disable")
+        assert resp.status_code == 200, resp.text
+        body = resp.json()
+        assert body["skill_name"] == "e2e_test_skill"
+        assert body["is_disabled"] is True
+
+        # 3. Verify via GET /skills that the skill is excluded.
+        resp = admin_client.get("/api/v1/skills")
+        assert resp.status_code == 200
+        names = [s["name"] for s in resp.json()]
+        assert "e2e_test_skill" not in names
+
+        # 4. POST /admin/skills/{name}/enable → enabled.
+        resp = admin_client.post("/api/v1/admin/skills/e2e_test_skill/enable")
+        assert resp.status_code == 200
+        assert resp.json()["enabled"] is True
+
+        # Verify via GET /skills that the skill is included again.
+        resp = admin_client.get("/api/v1/skills")
+        assert resp.status_code == 200
+        names = [s["name"] for s in resp.json()]
+        assert "e2e_test_skill" in names
+
+    def test_usage_dashboard_flow(self, admin_client: TestClient):
+        """Query usage endpoints with empty data → verify empty results."""
+        # The admin_app fixture doesn't install an llm_gateway on app.state,
+        # so the usage routes will return 500. We need to install a stub
+        # gateway with an empty InMemoryUsageStore.
+        from agentkit.llm.providers.usage_store import InMemoryUsageStore
+
+        class _StubTracker:
+            def __init__(self, store):
+                self.store = store
+
+        class _StubGateway:
+            def __init__(self, store):
+                self._usage_tracker = _StubTracker(store)
+
+        store = InMemoryUsageStore()
+        # Install the stub gateway on the app state.
+        admin_client.app.state.llm_gateway = _StubGateway(store)
+
+        # 1. GET /admin/usage/summary → 200 with zeros.
+        resp = admin_client.get("/api/v1/admin/usage/summary")
+        assert resp.status_code == 200
+        body = resp.json()
+        assert body["total_tokens"] == 0
+        assert body["total_cost"] == 0.0
+        assert body["total_requests"] == 0
+
+        # 2. GET /admin/usage/timeseries → 200 with empty list.
+        resp = admin_client.get("/api/v1/admin/usage/timeseries")
+        assert resp.status_code == 200
+        assert resp.json() == []
+
+        # 3. GET /admin/usage/by-model → 200 with empty list.
+        resp = admin_client.get("/api/v1/admin/usage/by-model")
+        assert resp.status_code == 200
+        assert resp.json() == []
+
+        # 4. GET /admin/usage/top-users → 200 with empty list.
+        resp = admin_client.get("/api/v1/admin/usage/top-users")
+        assert resp.status_code == 200
+        assert resp.json() == []
+
+        # 5. GET /admin/usage/export?format=csv → 200 with CSV header.
+        resp = admin_client.get("/api/v1/admin/usage/export", params={"format": "csv"})
+        assert resp.status_code == 200
+        assert resp.headers["content-type"].startswith("text/csv")
+        reader = csv.DictReader(io.StringIO(resp.text))
+        rows = list(reader)
+        assert rows == []
+        # Header should still be present.
+        assert "timestamp" in resp.text
+        assert "user_id" in resp.text
+        assert "department_id" in resp.text
diff --git a/tests/integration/admin/test_quota_enforcement.py b/tests/integration/admin/test_quota_enforcement.py
new file mode 100644
index 0000000..d79305a
--- /dev/null
+++ b/tests/integration/admin/test_quota_enforcement.py
@@ -0,0 +1,425 @@
+"""Quota enforcement integration tests (U10).
+
+Verifies quota enforcement end-to-end through the LLMGateway:
+
+- Token limit exceeded → QuotaExceededError raised.
+- Cost limit exceeded → QuotaExceededError raised.
+- Model not in whitelist → QuotaExceededError raised.
+- No quota set → request allowed.
+- Multi-department: strictest-wins (one exceeds, other doesn't → rejected).
+- Integration test with real LLMGateway + mock provider + InMemoryUsageStore.
+
+These tests use a real :class:`LLMGateway` with a :class:`FakeProvider`
+(mock LLM provider) and a real :class:`QuotaService` backed by a temp
+SQLite auth DB. No external services (Redis, real LLM API) are required.
+"""
+
+from __future__ import annotations
+
+import uuid
+from pathlib import Path
+
+import pytest
+
+from agentkit.llm.gateway import LLMGateway, QuotaExceededError
+from agentkit.llm.protocol import (
+    LLMProvider,
+    LLMRequest,
+    LLMResponse,
+    TokenUsage,
+)
+from agentkit.llm.providers.usage_store import InMemoryUsageStore
+from agentkit.server.admin.quota_service import (
+    get_quota_service,
+    set_quota_service,
+)
+from agentkit.server.auth.models import init_auth_db
+
+
+# ---------------------------------------------------------------------------
+# Test doubles
+# ---------------------------------------------------------------------------
+
+
+class FakeProvider(LLMProvider):
+    """A minimal LLMProvider that returns a fixed response.
+
+    The response usage (prompt_tokens, completion_tokens) can be
+    customized per-instance to simulate different token consumption.
+    """
+
+    def __init__(
+        self,
+        name: str = "fake",
+        prompt_tokens: int = 100,
+        completion_tokens: int = 50,
+    ) -> None:
+        self._name = name
+        self._prompt_tokens = prompt_tokens
+        self._completion_tokens = completion_tokens
+        self.last_request: LLMRequest | None = None
+        self.call_count = 0
+
+    async def chat(self, request: LLMRequest) -> LLMResponse:
+        self.last_request = request
+        self.call_count += 1
+        return LLMResponse(
+            content=f"response from {self._name}",
+            model=request.model,
+            usage=TokenUsage(
+                prompt_tokens=self._prompt_tokens,
+                completion_tokens=self._completion_tokens,
+            ),
+        )
+
+
+# ---------------------------------------------------------------------------
+# Fixtures
+# ---------------------------------------------------------------------------
+
+
+@pytest.fixture
+def store() -> InMemoryUsageStore:
+    return InMemoryUsageStore()
+
+
+@pytest.fixture
+def gateway(store: InMemoryUsageStore) -> LLMGateway:
+    """A real LLMGateway with a FakeProvider registered as "openai"."""
+    gw = LLMGateway(usage_store=store)
+    gw.register_provider("openai", FakeProvider("openai"))
+    return gw
+
+
+@pytest.fixture
+async def fresh_db(tmp_path: Path) -> Path:
+    """A brand-new auth DB on a fresh path (no data)."""
+    db_path = tmp_path / "quota_enforcement.db"
+    await init_auth_db(db_path)
+    return db_path
+
+
+@pytest.fixture(autouse=True)
+def _reset_quota_singleton():
+    """Reset the QuotaService singleton before and after each test."""
+    set_quota_service(None)
+    yield
+    set_quota_service(None)
+
+
+def _random_dept_id() -> str:
+    return str(uuid.uuid4())
+
+
+# ---------------------------------------------------------------------------
+# Quota enforcement tests
+# ---------------------------------------------------------------------------
+
+
+class TestQuotaEnforcement:
+    """Tests for quota enforcement in LLM calls."""
+
+    async def test_token_limit_blocks_request(
+        self, gateway: LLMGateway, store: InMemoryUsageStore, fresh_db: Path
+    ):
+        """When department exceeds token limit, LLM call raises QuotaExceededError."""
+        dept_id = _random_dept_id()
+        svc = get_quota_service()
+        # Set a 100-token daily limit.
+        await svc.set_quota(fresh_db, dept_id, "token_limit", 100, period="daily")
+
+        # Pre-populate usage with 90 tokens (just under the limit).
+        gateway._usage_tracker.record(
+            agent_name="prev",
+            model="openai/gpt-4o",
+            usage=TokenUsage(prompt_tokens=60, completion_tokens=30),
+            cost=0.0,
+            latency_ms=10,
+            user_id="u1",
+            department_id=dept_id,
+        )
+
+        # The FakeProvider would use 100+50=150 tokens, but the quota
+        # check happens BEFORE the provider call. Since current usage
+        # (90) + nothing is checked — the gateway checks current_usage
+        # >= limit, which is 90 < 100, so this would actually pass.
+        #
+        # To force a block, we pre-populate usage AT the limit (100
+        # tokens). The check is `current_usage >= limit`, so 100 >= 100
+        # → blocked.
+        store._records.clear()
+        gateway._usage_tracker.record(
+            agent_name="prev",
+            model="openai/gpt-4o",
+            usage=TokenUsage(prompt_tokens=70, completion_tokens=30),
+            cost=0.0,
+            latency_ms=10,
+            user_id="u1",
+            department_id=dept_id,
+        )
+
+        with pytest.raises(QuotaExceededError) as exc_info:
+            await gateway.chat(
+                messages=[{"role": "user", "content": "hi"}],
+                model="openai/gpt-4o",
+                user_id="u1",
+                department_ids=[dept_id],
+                db_path=fresh_db,
+            )
+        err = exc_info.value
+        assert err.department_id == dept_id
+        assert err.quota_type == "token_limit"
+        assert err.period == "daily"
+        assert err.limit == 100
+        assert err.current == 100  # 70 prompt + 30 completion
+
+    async def test_cost_limit_blocks_request(
+        self, gateway: LLMGateway, store: InMemoryUsageStore, fresh_db: Path
+    ):
+        """When department exceeds cost limit, LLM call raises QuotaExceededError."""
+        dept_id = _random_dept_id()
+        svc = get_quota_service()
+        # cost_limit is in cents. Set 100 cents ($1.00) daily limit.
+        await svc.set_quota(fresh_db, dept_id, "cost_limit", 100, period="daily")
+
+        # Pre-populate usage with $1.50 cost = 150 cents, exceeding the
+        # 100-cent limit.
+        gateway._usage_tracker.record(
+            agent_name="prev",
+            model="openai/gpt-4o",
+            usage=TokenUsage(prompt_tokens=100, completion_tokens=50),
+            cost=1.50,  # $1.50 = 150 cents
+            latency_ms=10,
+            user_id="u1",
+            department_id=dept_id,
+        )
+
+        with pytest.raises(QuotaExceededError) as exc_info:
+            await gateway.chat(
+                messages=[{"role": "user", "content": "hi"}],
+                model="openai/gpt-4o",
+                user_id="u1",
+                department_ids=[dept_id],
+                db_path=fresh_db,
+            )
+        err = exc_info.value
+        assert err.quota_type == "cost_limit"
+        assert err.period == "daily"
+        assert err.limit == 100
+        # current is in cents (150 cents = $1.50).
+        assert err.current == 150.0
+
+    async def test_model_whitelist_blocks_unlisted_model(self, gateway: LLMGateway, fresh_db: Path):
+        """When model not in whitelist, LLM call is rejected."""
+        dept_id = _random_dept_id()
+        svc = get_quota_service()
+        # Whitelist only allows "claude" — gateway is calling "gpt-4o".
+        await svc.set_quota(fresh_db, dept_id, "model_whitelist", ["claude"], period="daily")
+
+        with pytest.raises(QuotaExceededError) as exc_info:
+            await gateway.chat(
+                messages=[{"role": "user", "content": "hi"}],
+                model="openai/gpt-4o",
+                user_id="u1",
+                department_ids=[dept_id],
+                db_path=fresh_db,
+            )
+        err = exc_info.value
+        assert err.quota_type == "model_whitelist"
+        assert err.department_id == dept_id
+        # For model_whitelist, current is the rejected model name.
+        assert err.current == "openai/gpt-4o"
+
+    async def test_no_quota_allows_all(self, gateway: LLMGateway, fresh_db: Path):
+        """Without any quota set, all requests are allowed."""
+        dept_id = _random_dept_id()
+        # No quota set — request should succeed.
+        response = await gateway.chat(
+            messages=[{"role": "user", "content": "hi"}],
+            model="openai/gpt-4o",
+            user_id="u1",
+            department_ids=[dept_id],
+            db_path=fresh_db,
+        )
+        assert response.content == "response from openai"
+        assert response.usage.total_tokens == 150  # 100 prompt + 50 completion
+
+    async def test_multi_department_strictest_wins(
+        self, gateway: LLMGateway, store: InMemoryUsageStore, fresh_db: Path
+    ):
+        """User in depts A+B: A has quota, B doesn't → A's quota applies.
+
+        Strictest-wins: if ANY department fails ANY check, the request
+        is rejected.
+        """
+        dept_a = _random_dept_id()
+        dept_b = _random_dept_id()
+        svc = get_quota_service()
+        # Set a 1-token limit on dept A only; dept B has no quota.
+        await svc.set_quota(fresh_db, dept_a, "token_limit", 1, period="daily")
+
+        # Pre-populate usage for dept A so it exceeds the 1-token limit.
+        gateway._usage_tracker.record(
+            agent_name="prev",
+            model="openai/gpt-4o",
+            usage=TokenUsage(prompt_tokens=10, completion_tokens=5),
+            cost=0.0,
+            latency_ms=10,
+            user_id="u1",
+            department_id=dept_a,
+        )
+
+        with pytest.raises(QuotaExceededError) as exc_info:
+            await gateway.chat(
+                messages=[{"role": "user", "content": "hi"}],
+                model="openai/gpt-4o",
+                user_id="u1",
+                department_ids=[dept_a, dept_b],
+                db_path=fresh_db,
+            )
+        # The error should reference dept_a (the one that exceeded).
+        assert exc_info.value.department_id == dept_a
+        assert exc_info.value.quota_type == "token_limit"
+
+    async def test_quota_check_with_real_gateway(
+        self, gateway: LLMGateway, store: InMemoryUsageStore, fresh_db: Path
+    ):
+        """Integration test with real LLMGateway + mock provider.
+
+        Verifies the full flow:
+        1. Quota check happens before the provider call.
+        2. On success, usage is recorded with the correct department_id.
+        3. The usage record carries user_id + department_id.
+        """
+        dept_id = _random_dept_id()
+        svc = get_quota_service()
+        # Set a generous token limit (1M tokens) — should not block.
+        await svc.set_quota(fresh_db, dept_id, "token_limit", 1_000_000, period="daily")
+
+        # Make the LLM call.
+        response = await gateway.chat(
+            messages=[{"role": "user", "content": "hi"}],
+            model="openai/gpt-4o",
+            user_id="u1",
+            department_ids=[dept_id],
+            db_path=fresh_db,
+        )
+        assert response.content == "response from openai"
+
+        # Verify usage was recorded with the correct attributes.
+        summary = store.get_usage()
+        assert len(summary.records) == 1
+        rec = summary.records[0]
+        assert rec.user_id == "u1"
+        assert rec.department_id == dept_id
+        assert rec.model == "gpt-4o"
+        assert rec.total_tokens == 150  # 100 prompt + 50 completion
+
+        # Verify the quota check counted this usage (next call should
+        # still pass since the limit is 1M tokens).
+        response2 = await gateway.chat(
+            messages=[{"role": "user", "content": "hi again"}],
+            model="openai/gpt-4o",
+            user_id="u1",
+            department_ids=[dept_id],
+            db_path=fresh_db,
+        )
+        assert response2.content == "response from openai"
+
+        # Now there should be 2 usage records.
+        summary = store.get_usage()
+        assert len(summary.records) == 2
+        # All records should carry the department_id.
+        assert all(r.department_id == dept_id for r in summary.records)
+
+    async def test_quota_check_skipped_without_db_path(self, gateway: LLMGateway, fresh_db: Path):
+        """When db_path is None, no quota check is performed."""
+        dept_id = _random_dept_id()
+        svc = get_quota_service()
+        # Set a tiny quota that would normally block.
+        await svc.set_quota(fresh_db, dept_id, "token_limit", 1, period="daily")
+
+        # Call without db_path — should succeed (no quota check).
+        response = await gateway.chat(
+            messages=[{"role": "user", "content": "hi"}],
+            model="openai/gpt-4o",
+            user_id="u1",
+            department_ids=[dept_id],
+            db_path=None,
+        )
+        assert response.content == "response from openai"
+
+    async def test_quota_check_skipped_without_department_ids(
+        self, gateway: LLMGateway, fresh_db: Path
+    ):
+        """When department_ids is None, no quota check is performed."""
+        response = await gateway.chat(
+            messages=[{"role": "user", "content": "hi"}],
+            model="openai/gpt-4o",
+            user_id="u1",
+            department_ids=None,
+            db_path=fresh_db,
+        )
+        assert response.content == "response from openai"
+
+    async def test_model_whitelist_allows_listed_model(self, gateway: LLMGateway, fresh_db: Path):
+        """Model in whitelist → request allowed."""
+        dept_id = _random_dept_id()
+        svc = get_quota_service()
+        # Whitelist uses the full resolved model identifier (provider/model).
+        await svc.set_quota(
+            fresh_db,
+            dept_id,
+            "model_whitelist",
+            ["openai/gpt-4o"],
+            period="daily",
+        )
+        response = await gateway.chat(
+            messages=[{"role": "user", "content": "hi"}],
+            model="openai/gpt-4o",
+            user_id="u1",
+            department_ids=[dept_id],
+            db_path=fresh_db,
+        )
+        assert response.content == "response from openai"
+
+    async def test_quota_check_uses_correct_period_window(
+        self, gateway: LLMGateway, store: InMemoryUsageStore, fresh_db: Path
+    ):
+        """Quota check uses the daily window (since 00:00 UTC today).
+
+        The quota check happens BEFORE the LLM call, using the current
+        accumulated usage. So:
+        - 1st call: usage=0, check 0 >= 150 → False → allowed. After
+          the call, usage=150.
+        - 2nd call: usage=150, check 150 >= 150 → True → blocked.
+        """
+        dept_id = _random_dept_id()
+        svc = get_quota_service()
+        # Set a 150-token daily limit (the FakeProvider uses 150 tokens
+        # per call: 100 prompt + 50 completion).
+        await svc.set_quota(fresh_db, dept_id, "token_limit", 150, period="daily")
+
+        # First call: current usage is 0, under the 150 limit → allowed.
+        response = await gateway.chat(
+            messages=[{"role": "user", "content": "hi"}],
+            model="openai/gpt-4o",
+            user_id="u1",
+            department_ids=[dept_id],
+            db_path=fresh_db,
+        )
+        assert response.content == "response from openai"
+
+        # Second call: current usage is now 150 (from the first call),
+        # which is >= the 150-token limit → blocked.
+        with pytest.raises(QuotaExceededError) as exc_info:
+            await gateway.chat(
+                messages=[{"role": "user", "content": "hi again"}],
+                model="openai/gpt-4o",
+                user_id="u1",
+                department_ids=[dept_id],
+                db_path=fresh_db,
+            )
+        assert exc_info.value.quota_type == "token_limit"
+        assert exc_info.value.current == 150  # accumulated from first call
+        assert exc_info.value.limit == 150
diff --git a/tests/integration/admin/test_security_isolation.py b/tests/integration/admin/test_security_isolation.py
new file mode 100644
index 0000000..bd2f076
--- /dev/null
+++ b/tests/integration/admin/test_security_isolation.py
@@ -0,0 +1,483 @@
+"""Security isolation integration tests for department-scoped resources (U10).
+
+Verifies department-based access control end-to-end through the full
+request stack (route → DepartmentContext → filtering → response):
+
+- User in dept A cannot see skills/KB bound to dept B.
+- User in depts A+B sees the union of both departments' resources.
+- Admin sees all resources regardless of department bindings.
+- User removed from a department loses access to that department's
+  resources.
+- Disabled department's resources are not visible to its users.
+- API-key client (no user_id) sees only global resources.
+- Non-admin user gets 403 on all admin endpoints.
+
+The tests mount a minimal FastAPI app with the ``skills`` and
+``kb-management`` routers (for isolation verification) plus the
+``admin_router`` (for the 403 checks). The ``get_department_context``
+dependency is overridden per-test to simulate different callers.
+"""
+
+from __future__ import annotations
+
+import sqlite3
+import uuid
+from datetime import datetime, timezone
+from pathlib import Path
+from typing import Any
+
+import pytest
+from fastapi import FastAPI, HTTPException, Request
+from fastapi.testclient import TestClient
+
+from agentkit.server.admin.context import DepartmentContext
+from agentkit.server.auth.models import init_auth_db
+from agentkit.server.routes import admin as admin_routes_module
+from agentkit.server.routes import kb_management as kb_routes
+from agentkit.server.routes import skills as skills_routes
+from agentkit.skills.base import Skill, SkillConfig
+from agentkit.skills.registry import SkillRegistry
+
+
+# ---------------------------------------------------------------------------
+# Fixtures
+# ---------------------------------------------------------------------------
+
+
+@pytest.fixture
+async def tmp_auth_db(tmp_path: Path) -> Path:
+    db_path = tmp_path / "security_isolation.db"
+    await init_auth_db(db_path)
+    return db_path
+
+
+@pytest.fixture
+def skill_registry() -> SkillRegistry:
+    """A SkillRegistry pre-loaded with three test skills.
+
+    - ``hr_skill`` — will be bound to department A.
+    - ``eng_skill`` — will be bound to department B.
+    - ``global_skill`` — has NO department binding (global).
+    """
+    registry = SkillRegistry()
+    for name in ("hr_skill", "eng_skill", "global_skill"):
+        config = SkillConfig(
+            name=name,
+            agent_type="test_type",
+            task_mode="llm_generate",
+            description=f"Test skill {name}",
+            prompt={"identity": name, "instructions": "test"},
+        )
+        registry.register(Skill(config=config))
+    return registry
+
+
+@pytest.fixture
+def kb_store():
+    """Reset the module-level KB source store singleton."""
+    kb_routes._source_store = kb_routes.KnowledgeSourceStore()
+    return kb_routes._source_store
+
+
+@pytest.fixture
+def app(
+    tmp_auth_db: Path,
+    skill_registry: SkillRegistry,
+    kb_store: kb_routes.KnowledgeSourceStore,
+) -> FastAPI:
+    """A FastAPI app with skills + kb-management + admin routers mounted.
+
+    The ``get_department_context`` dependency is overridden per-test via
+    ``app.dependency_overrides``. The default override is
+    "unauthenticated caller".
+    """
+    application = FastAPI()
+    application.state.auth_db_path = str(tmp_auth_db)
+    application.state.skill_registry = skill_registry
+
+    application.include_router(skills_routes.router, prefix="/api/v1")
+    application.include_router(kb_routes.router, prefix="/api/v1")
+    application.include_router(admin_routes_module.admin_router, prefix="/api/v1")
+
+    # Default: unauthenticated caller.
+    application.dependency_overrides[skills_routes.get_department_context] = (
+        _unauthenticated_context
+    )
+    application.dependency_overrides[kb_routes.get_department_context] = _unauthenticated_context
+    # Default: admin access allowed (used only for the admin 403 test
+    # which overrides this).
+    application.dependency_overrides[admin_routes_module._require_admin] = lambda: (
+        _make_admin_user()
+    )
+    return application
+
+
+@pytest.fixture
+def client(app: FastAPI) -> TestClient:
+    return TestClient(app)
+
+
+# ---------------------------------------------------------------------------
+# Department-context dependency overrides
+# ---------------------------------------------------------------------------
+
+
+async def _unauthenticated_context(request: Request) -> DepartmentContext:
+    """Simulate an unauthenticated caller (no current_user)."""
+    return DepartmentContext(user_id=None, department_ids=[], is_admin=False)
+
+
+def _ctx_for_user(
+    user_id: str | None,
+    department_ids: list[str],
+    is_admin: bool = False,
+):
+    """Build a dependency override returning a fixed DepartmentContext."""
+
+    async def _override(request: Request) -> DepartmentContext:
+        return DepartmentContext(
+            user_id=user_id,
+            department_ids=list(department_ids),
+            is_admin=is_admin,
+        )
+
+    return _override
+
+
+def _set_caller(
+    app: FastAPI,
+    user_id: str | None,
+    department_ids: list[str],
+    is_admin: bool = False,
+) -> None:
+    """Install the dependency overrides for both skills + kb routers."""
+    override = _ctx_for_user(user_id, department_ids, is_admin)
+    app.dependency_overrides[skills_routes.get_department_context] = override
+    app.dependency_overrides[kb_routes.get_department_context] = override
+
+
+def _make_admin_user() -> dict[str, Any]:
+    return {"user_id": "admin-1", "username": "admin", "role": "admin"}
+
+
+def _raise_forbidden() -> dict[str, Any]:
+    raise HTTPException(status_code=403, detail="Admin permission required")
+
+
+# ---------------------------------------------------------------------------
+# DB helpers (synchronous sqlite3 — no event-loop mixing with TestClient)
+# ---------------------------------------------------------------------------
+
+
+def _now_iso() -> str:
+    return datetime.now(timezone.utc).isoformat()
+
+
+def _create_department(db_path: Path, name: str) -> str:
+    dept_id = str(uuid.uuid4())
+    with sqlite3.connect(str(db_path)) as db:
+        db.execute(
+            "INSERT INTO departments (id, name, description, is_active, created_at) "
+            "VALUES (?, ?, ?, ?, ?)",
+            (dept_id, name, "", 1, _now_iso()),
+        )
+        db.commit()
+    return dept_id
+
+
+def _disable_department(db_path: Path, department_id: str) -> None:
+    with sqlite3.connect(str(db_path)) as db:
+        db.execute(
+            "UPDATE departments SET is_active = 0 WHERE id = ?",
+            (department_id,),
+        )
+        db.commit()
+
+
+def _bind_skill(db_path: Path, department_id: str, skill_name: str) -> None:
+    with sqlite3.connect(str(db_path)) as db:
+        db.execute(
+            "INSERT INTO department_skill_bindings (id, department_id, skill_name, created_at) "
+            "VALUES (?, ?, ?, ?)",
+            (str(uuid.uuid4()), department_id, skill_name, _now_iso()),
+        )
+        db.commit()
+
+
+def _bind_kb(db_path: Path, department_id: str, kb_source_id: str) -> None:
+    with sqlite3.connect(str(db_path)) as db:
+        db.execute(
+            "INSERT INTO department_kb_bindings (id, department_id, kb_source_id, created_at) "
+            "VALUES (?, ?, ?, ?)",
+            (str(uuid.uuid4()), department_id, kb_source_id, _now_iso()),
+        )
+        db.commit()
+
+
+def _assign_user_to_department(db_path: Path, user_id: str, department_id: str) -> None:
+    with sqlite3.connect(str(db_path)) as db:
+        db.execute(
+            "INSERT INTO user_departments (user_id, department_id, created_at) VALUES (?, ?, ?)",
+            (user_id, department_id, _now_iso()),
+        )
+        db.commit()
+
+
+def _remove_user_from_department(db_path: Path, user_id: str, department_id: str) -> None:
+    with sqlite3.connect(str(db_path)) as db:
+        db.execute(
+            "DELETE FROM user_departments WHERE user_id = ? AND department_id = ?",
+            (user_id, department_id),
+        )
+        db.commit()
+
+
+# ---------------------------------------------------------------------------
+# Test fixture: departments A and B with skill/KB bindings
+# ---------------------------------------------------------------------------
+
+
+@pytest.fixture
+def dept_setup(tmp_auth_db: Path, kb_store: kb_routes.KnowledgeSourceStore):
+    """Create departments A and B, bind skills and KB sources.
+
+    Layout:
+      - Department A: bound to ``hr_skill`` and KB source ``hr_kb``
+      - Department B: bound to ``eng_skill`` and KB source ``eng_kb``
+      - ``global_skill`` and ``global_kb`` have NO bindings (global)
+    """
+    dept_a = _create_department(tmp_auth_db, "HR")
+    dept_b = _create_department(tmp_auth_db, "Engineering")
+    _bind_skill(tmp_auth_db, dept_a, "hr_skill")
+    _bind_skill(tmp_auth_db, dept_b, "eng_skill")
+    # global_skill intentionally has no binding.
+
+    # Create KB sources in the in-memory store.
+    hr_kb = kb_store.add_source("HR KB", "local", {})
+    eng_kb = kb_store.add_source("Engineering KB", "local", {})
+    global_kb = kb_store.add_source("Global KB", "local", {})
+    _bind_kb(tmp_auth_db, dept_a, hr_kb.id)
+    _bind_kb(tmp_auth_db, dept_b, eng_kb.id)
+    # global_kb intentionally has no binding.
+
+    return {
+        "dept_a": dept_a,
+        "dept_b": dept_b,
+        "hr_kb_id": hr_kb.id,
+        "eng_kb_id": eng_kb.id,
+        "global_kb_id": global_kb.id,
+    }
+
+
+# ---------------------------------------------------------------------------
+# Department isolation tests
+# ---------------------------------------------------------------------------
+
+
+class TestDepartmentIsolation:
+    """Security tests for department-based resource isolation."""
+
+    def test_user_cannot_see_other_department_skills(
+        self, app: FastAPI, client: TestClient, dept_setup: dict
+    ):
+        """User in dept A cannot see skills bound to dept B."""
+        _set_caller(app, user_id="alice", department_ids=[dept_setup["dept_a"]])
+        resp = client.get("/api/v1/skills")
+        assert resp.status_code == 200
+        names = {s["name"] for s in resp.json()}
+        # Alice sees hr_skill (dept A) + global_skill, NOT eng_skill (dept B).
+        assert "hr_skill" in names
+        assert "global_skill" in names
+        assert "eng_skill" not in names
+
+    def test_user_cannot_see_other_department_kb(
+        self, app: FastAPI, client: TestClient, dept_setup: dict
+    ):
+        """User in dept A cannot see KB sources bound to dept B."""
+        _set_caller(app, user_id="alice", department_ids=[dept_setup["dept_a"]])
+        resp = client.get("/api/v1/kb-management/sources")
+        assert resp.status_code == 200
+        ids = {s["id"] for s in resp.json()["sources"]}
+        # Alice sees hr_kb (dept A) + global_kb, NOT eng_kb (dept B).
+        assert dept_setup["hr_kb_id"] in ids
+        assert dept_setup["global_kb_id"] in ids
+        assert dept_setup["eng_kb_id"] not in ids
+
+    def test_user_in_multiple_departments_sees_union(
+        self, app: FastAPI, client: TestClient, dept_setup: dict
+    ):
+        """User in depts A+B sees skills from both."""
+        _set_caller(
+            app,
+            user_id="alice",
+            department_ids=[dept_setup["dept_a"], dept_setup["dept_b"]],
+        )
+        resp = client.get("/api/v1/skills")
+        assert resp.status_code == 200
+        names = {s["name"] for s in resp.json()}
+        # Alice sees hr_skill (A) + eng_skill (B) + global_skill.
+        assert names == {"hr_skill", "eng_skill", "global_skill"}
+
+    def test_admin_sees_all_resources(self, app: FastAPI, client: TestClient, dept_setup: dict):
+        """Admin user sees all resources regardless of department bindings."""
+        _set_caller(
+            app,
+            user_id="admin-1",
+            department_ids=[],
+            is_admin=True,
+        )
+        # Admin sees all skills.
+        resp = client.get("/api/v1/skills")
+        assert resp.status_code == 200
+        names = {s["name"] for s in resp.json()}
+        assert names == {"hr_skill", "eng_skill", "global_skill"}
+
+        # Admin sees all KB sources.
+        resp = client.get("/api/v1/kb-management/sources")
+        assert resp.status_code == 200
+        ids = {s["id"] for s in resp.json()["sources"]}
+        assert ids == {
+            dept_setup["hr_kb_id"],
+            dept_setup["eng_kb_id"],
+            dept_setup["global_kb_id"],
+        }
+
+    def test_user_removed_from_department_loses_access(
+        self,
+        app: FastAPI,
+        client: TestClient,
+        tmp_auth_db: Path,
+        dept_setup: dict,
+    ):
+        """User removed from dept A can no longer see dept A's skills."""
+        user_id = "user-removal"
+        dept_a = dept_setup["dept_a"]
+
+        # Initially assign to dept A.
+        _assign_user_to_department(tmp_auth_db, user_id, dept_a)
+        _set_caller(app, user_id=user_id, department_ids=[dept_a])
+
+        resp = client.get("/api/v1/skills")
+        names = {s["name"] for s in resp.json()}
+        assert "hr_skill" in names
+
+        # Remove from dept A — simulate the context change.
+        _remove_user_from_department(tmp_auth_db, user_id, dept_a)
+        _set_caller(app, user_id=user_id, department_ids=[])
+
+        resp = client.get("/api/v1/skills")
+        names = {s["name"] for s in resp.json()}
+        assert "hr_skill" not in names
+        assert "global_skill" in names
+
+    def test_disabled_department_excluded(
+        self,
+        app: FastAPI,
+        client: TestClient,
+        tmp_auth_db: Path,
+        dept_setup: dict,
+    ):
+        """Disabled department's resources are not visible to its users.
+
+        The ``_fetch_user_department_ids`` helper in
+        :mod:`agentkit.server.admin.context` filters out disabled
+        departments (``is_active=0``). We simulate this by disabling
+        dept A in the DB and updating the caller's context to reflect
+        the now-empty department list.
+        """
+        user_id = "user-disabled-dept"
+        dept_a = dept_setup["dept_a"]
+
+        # Initially assign to dept A and verify access.
+        _assign_user_to_department(tmp_auth_db, user_id, dept_a)
+        _set_caller(app, user_id=user_id, department_ids=[dept_a])
+        resp = client.get("/api/v1/skills")
+        names = {s["name"] for s in resp.json()}
+        assert "hr_skill" in names
+
+        # Disable dept A in the DB.
+        _disable_department(tmp_auth_db, dept_a)
+        # Simulate the context change: the next request's
+        # ``get_department_context`` would re-query user_departments
+        # and find no *active* departments.
+        _set_caller(app, user_id=user_id, department_ids=[])
+
+        resp = client.get("/api/v1/skills")
+        names = {s["name"] for s in resp.json()}
+        assert "hr_skill" not in names
+        assert "global_skill" in names
+
+    def test_api_key_client_sees_only_global(
+        self, app: FastAPI, client: TestClient, dept_setup: dict
+    ):
+        """API key client (no user_id) sees only global resources."""
+        # API-key client → user_id=None, department_ids=[], is_admin=False.
+        _set_caller(app, user_id=None, department_ids=[], is_admin=False)
+
+        # Skills: only global_skill.
+        resp = client.get("/api/v1/skills")
+        assert resp.status_code == 200
+        names = {s["name"] for s in resp.json()}
+        assert names == {"global_skill"}
+
+        # KB sources: only global_kb.
+        resp = client.get("/api/v1/kb-management/sources")
+        assert resp.status_code == 200
+        ids = {s["id"] for s in resp.json()["sources"]}
+        assert ids == {dept_setup["global_kb_id"]}
+
+
+# ---------------------------------------------------------------------------
+# Non-admin access tests
+# ---------------------------------------------------------------------------
+
+
+class TestNonAdminAccess:
+    """Non-admin user gets 403 on all admin endpoints."""
+
+    def test_non_admin_cannot_access_admin_endpoints(self, app: FastAPI, tmp_auth_db: Path):
+        """Non-admin user gets 403 on a representative sample of admin endpoints."""
+        # Override _require_admin to raise 403 (simulating a non-admin caller).
+        app.dependency_overrides[admin_routes_module._require_admin] = _raise_forbidden
+        client = TestClient(app)
+
+        # Representative sample of admin endpoints across all resource types.
+        endpoints = [
+            ("GET", "/api/v1/admin/departments"),
+            ("POST", "/api/v1/admin/departments"),
+            ("GET", "/api/v1/admin/users"),
+            ("POST", "/api/v1/admin/users"),
+            ("GET", "/api/v1/admin/llm/providers"),
+            ("POST", "/api/v1/admin/llm/providers"),
+            ("GET", "/api/v1/admin/usage/summary"),
+            ("GET", "/api/v1/admin/usage/timeseries"),
+            ("GET", "/api/v1/admin/usage/by-model"),
+            ("GET", "/api/v1/admin/usage/top-users"),
+            ("GET", "/api/v1/admin/usage/export"),
+        ]
+
+        for method, path in endpoints:
+            if method == "GET":
+                resp = client.get(path)
+            elif method == "POST":
+                # Use a minimal body where required; we expect 403 before
+                # body validation runs.
+                body: dict[str, Any] | None = {}
+                if path == "/api/v1/admin/departments":
+                    body = {"name": "X"}
+                elif path == "/api/v1/admin/users":
+                    body = {
+                        "username": "x",
+                        "email": "x@x.com",
+                        "password": "Pw123!",
+                    }
+                elif path == "/api/v1/admin/llm/providers":
+                    body = {
+                        "name": "x",
+                        "type": "openai",
+                        "api_key": "sk-x",
+                    }
+                resp = client.post(path, json=body)
+            assert resp.status_code == 403, (
+                f"{method} {path} returned {resp.status_code}, expected 403"
+            )