diging · Girik1105 · Mar 26, 2026
diff --git a/.app_env_example b/.app_env_example
@@ -13,6 +13,10 @@ source POSTGRES_PASSWORD=postgres
 
 source APP_ROOT = hopper/
 
+# MCP Knowledge Base Configuration
+source KB_MCP_HOST=http://host.docker.internal:8002
+source KB_MCP_JWT_TOKEN=your-jwt-token  # JWT token from OIDC token endpoint
+
 # JWT config
 source IDP_OIDC_PRIVATE_KEY="-----BEGIN PRIVATE KEY----- \
 <paste key here> \

diff --git a/hospexplorer/ask/kb_connector.py b/hospexplorer/ask/kb_connector.py
@@ -0,0 +1,30 @@
+import logging
+
+import httpx
+from django.conf import settings
+
+logger = logging.getLogger(__name__)
+
+
+def list_kb_documents(page=1, page_size=10):
+    """Call MCP KB docs/list endpoint. Returns parsed JSON response.
+
+    Response format: {total, page, page_size, documents: [{id, title, url, chunks: [...]}]}
+    """
+    headers = {
+        "Authorization": f"Bearer {settings.KB_MCP_JWT_TOKEN}",
+        "Content-Type": "application/json",
+    }
+    url = f"{settings.KB_MCP_HOST}/docs/list"
+    params = {"page": page, "page_size": page_size}
+
+    with httpx.Client() as client:
+        response = client.get(
+            url,
+            headers=headers,
+            params=params,
+            timeout=settings.KB_MCP_TIMEOUT,
+        )
+
+    response.raise_for_status()
+    return response.json()
diff --git a/hospexplorer/ask/templates/_base.html b/hospexplorer/ask/templates/_base.html
@@ -53,6 +53,16 @@
                     </div>
                 </div>
                 {% if user.is_authenticated %}
+                <div class="border-top p-2">
+                    <a href="{% url 'ask:kb-resources' %}" class="btn btn-outline-secondary btn-sm w-100 text-start">
+                        <svg xmlns="http://www.w3.org/2000/svg" width="14" height="14" fill="currentColor" class="me-1 mb-1" viewBox="0 0 16 16">
+                            <path d="M1 2.828c.885-.37 2.154-.769 3.388-.893 1.33-.134 2.458.063 3.112.752v9.746c-.935-.53-2.12-.603-3.213-.493-1.18.12-2.37.461-3.287.811V2.828zm7.5-.141c.654-.689 1.782-.886 3.112-.752 1.234.124 2.503.523 3.388.893v9.923c-.918-.35-2.107-.692-3.287-.81-1.094-.111-2.278-.039-3.213.492V2.687zM8 1.783C7.015.936 5.587.81 4.287.94c-1.514.153-3.042.672-3.994 1.105A.5.5 0 0 0 0 2.5v11a.5.5 0 0 0 .707.455c.882-.4 2.303-.881 3.68-1.02 1.409-.142 2.59.087 3.223.877a.5.5 0 0 0 .78 0c.633-.79 1.814-1.019 3.222-.877 1.378.139 2.8.62 3.681 1.02A.5.5 0 0 0 16 13.5v-11a.5.5 0 0 0-.293-.455c-.952-.433-2.48-.952-3.994-1.105C10.413.809 8.985.936 8 1.783z"/>
+                        </svg>
+                        Knowledge Base
+                    </a>
+                </div>
+                {% endif %}
+                {% if user.is_authenticated %}
                 <!-- User profile at bottom -->
                 <div class="border-top p-3 mt-auto dropup">
                     <a href="#" class="d-flex align-items-center text-decoration-none"

diff --git a/hospexplorer/ask/templates/kb/resources.html b/hospexplorer/ask/templates/kb/resources.html
@@ -0,0 +1,187 @@
+{% extends "_base.html" %}
+
+{% block content %}
+<div class="mt-4 mb-4" x-data="kbSync()">
+    <a href="{% url 'ask:index' %}" class="text-decoration-none mb-3 d-inline-block" style="color: rgb(197, 66, 130);">
+        &larr; Back to Chat
+    </a>
+    <div class="d-flex justify-content-between align-items-center mb-3">
+        <h2 class="mb-0">Knowledge Base Resources</h2>
+        <button @click="sync()" :disabled="syncing" class="btn text-white" style="background-color: #8C1D40;">
+            <span x-show="!syncing">Sync with Knowledge Base</span>
+            <span x-show="syncing">
+                <span class="spinner-border spinner-border-sm" role="status" aria-hidden="true"></span>
+                Syncing...
+            </span>
+        </button>
+    </div>
+
+    <!-- Sync error -->
+    <div x-show="syncError" x-cloak class="alert alert-danger alert-dismissible fade show">
+        <span x-text="syncError"></span>
+        <button type="button" class="btn-close" @click="syncError = null"></button>
+    </div>
+
+    <!-- Sync summary -->
+    <div x-show="syncResult" x-cloak class="alert alert-info alert-dismissible fade show">
+        <strong>Sync Complete:</strong>
+        <span x-text="syncSummary"></span>
+        <button type="button" class="btn-close" @click="syncResult = null"></button>
+    </div>
+
+    <!-- Resources table -->
+    <div class="card">
+        <div class="table-responsive">
+            <table class="table table-hover mb-0">
+                <thead class="table-light">
+                    <tr>
+                        <th>Title</th>
+                        <th>URL</th>
+                        <th>Last Included</th>
+                        <th>KB Status</th>
+                    </tr>
+                </thead>
+                <tbody>
+                    {% for resource in page_obj %}
+                    <tr :class="getRowClass({{ resource.id }})">
+                        <td>{{ resource.title }}</td>
+                        <td>
+                            <a href="{{ resource.url }}" target="_blank" rel="noopener" class="text-truncate d-inline-block" style="max-width: 300px;">
+                                {{ resource.url|truncatechars:60 }}
+                            </a>
+                        </td>
+                        <td>{{ resource.modified_at|date:"N j, Y, P" }}</td>
+                        <td>
+                            <span x-text="getStatus({{ resource.id }})">--</span>
+                        </td>
+                    </tr>
+                    {% empty %}
+                    <tr>
+                        <td colspan="4" class="text-muted text-center py-4">No resources found in the internal database.</td>
+                    </tr>
+                    {% endfor %}
+                </tbody>
+            </table>
+        </div>
+    </div>
+
+    <!-- Untracked KB documents -->
+    <div x-show="untracked.length > 0" x-cloak class="mt-4">
+        <h5>Documents in KB but not tracked internally</h5>
+        <div class="card">
+            <div class="table-responsive">
+                <table class="table table-hover mb-0">
+                    <thead class="table-light">
+                        <tr>
+                            <th>Title</th>
+                            <th>URL</th>
+                        </tr>
+                    </thead>
+                    <tbody>
+                        <template x-for="doc in untracked" :key="doc.url">
+                            <tr class="table-info">
+                                <td x-text="doc.title"></td>
+                                <td>
+                                    <a :href="doc.url" target="_blank" rel="noopener" x-text="doc.url" class="text-truncate d-inline-block" style="max-width: 400px;"></a>
+                                </td>
+                            </tr>
+                        </template>
+                    </tbody>
+                </table>
+            </div>
+        </div>
+    </div>
+
+    <!-- Pagination -->
+    {% if page_obj.has_other_pages %}
+    <nav class="mt-3">
+        <ul class="pagination justify-content-center">
+            {% if page_obj.has_previous %}
+            <li class="page-item">
+                <a class="page-link" href="?page={{ page_obj.previous_page_number }}">Previous</a>
+            </li>
+            {% else %}
+            <li class="page-item disabled"><span class="page-link">Previous</span></li>
+            {% endif %}
+
+            {% for num in page_obj.paginator.page_range %}
+            <li class="page-item {% if page_obj.number == num %}active{% endif %}">
+                <a class="page-link" href="?page={{ num }}">{{ num }}</a>
+            </li>
+            {% endfor %}
+
+            {% if page_obj.has_next %}
+            <li class="page-item">
+                <a class="page-link" href="?page={{ page_obj.next_page_number }}">Next</a>
+            </li>
+            {% else %}
+            <li class="page-item disabled"><span class="page-link">Next</span></li>
+            {% endif %}
+        </ul>
+    </nav>
+    {% endif %}
+</div>
+
+<script>
+function kbSync() {
+    return {
+        syncing: false,
+        syncResult: null,
+        syncError: null,
+        statusMap: {},
+        untracked: [],
+        syncSummary: '',
+
+        async sync() {
+            this.syncing = true;
+            this.syncError = null;
+            this.syncResult = null;
+            this.untracked = [];
+            try {
+                const response = await fetch('{% url "ask:kb-sync" %}', {
+                    method: 'POST',
+                    headers: { 'X-CSRFToken': '{{ csrf_token }}' },
+                });
+                const data = await response.json();
+                if (data.success) {
+                    this.syncResult = data;
+                    this.statusMap = {};
+                    for (const r of data.resources) {
+                        this.statusMap[r.id] = r.status;
+                    }
+                    this.untracked = data.untracked || [];
+                    const inSync = data.resources.filter(r => r.status === 'in_kb').length;
+                    const missing = data.resources.filter(r => r.status === 'missing_from_kb').length;
+                    this.syncSummary = `${inSync} in sync, ${missing} missing from KB, ${this.untracked.length} in KB but untracked. (${data.kb_total} KB documents total)`;
+                } else {
+                    this.syncError = data.error || 'Sync failed.';
+                }
+            } catch (e) {
+                this.syncError = 'Failed to connect. Please try again.';
+            }
+            this.syncing = false;
+        },
+
+        getStatus(id) {
+            if (!this.syncResult) return '--';
+            const s = this.statusMap[id];
+            if (s === 'in_kb') return 'In Sync';
+            if (s === 'missing_from_kb') return 'Missing from KB';
+            return '--';
+        },
+
+        getRowClass(id) {
+            if (!this.syncResult) return '';
+            const s = this.statusMap[id];
+            if (s === 'in_kb') return 'table-success';
+            if (s === 'missing_from_kb') return 'table-warning';
+            return '';
+        }
+    };
+}
+</script>
+
+<style>
+[x-cloak] { display: none !important; }
+</style>
+{% endblock %}
diff --git a/hospexplorer/ask/urls.py b/hospexplorer/ask/urls.py
@@ -13,4 +13,6 @@
     re_path(r"^terms/$", views.terms_view, name="terms-view"),
     re_path(r"^terms/accept/$", views.terms_accept, name="terms-accept"),
     re_path(r"^history/delete$", views.delete_history, name="delete-history"),
+    re_path(r"^kb/$", views.kb_resources, name="kb-resources"),
+    re_path(r"^kb/sync/$", views.kb_sync, name="kb-sync"),
 ]
diff --git a/hospexplorer/ask/views.py b/hospexplorer/ask/views.py
@@ -8,8 +8,12 @@
 from django.contrib.auth.decorators import login_required
 from django.views.decorators.http import require_GET, require_POST, require_http_methods
 
-from ask.models import Conversation, QARecord, QueryTask, TermsAcceptance
+import httpx
+from django.core.paginator import Paginator
+
+from ask.models import Conversation, QARecord, QueryTask, TermsAcceptance, WebsiteResource
 from ask.tasks import run_llm_task
+from ask.kb_connector import list_kb_documents
 
 logger = logging.getLogger(__name__)
 
@@ -176,3 +180,90 @@ def terms_view(request):
 def delete_history(request):
     request.user.conversations.all().delete()
     return JsonResponse({"message": "All conversations deleted successfully!"})
+
+
+@login_required
+def kb_resources(request):
+    """Display paginated list of Knowledge Base resources from internal DB."""
+    resources = WebsiteResource.objects.all().order_by("-modified_at")
+    paginator = Paginator(resources, settings.KB_RESOURCES_PAGE_SIZE)
+    page_number = request.GET.get("page", 1)
+    page_obj = paginator.get_page(page_number)
+    return render(request, "kb/resources.html", {"page_obj": page_obj})
+
+
+@login_required
+@require_POST
+def kb_sync(request):
+    """Compare internal WebsiteResource records with MCP KB documents.
+
+    How it works:
+    1. Paginates through ALL documents from the MCP KB server via GET /docs/list
+       (each doc has: id, title, url, chunks). Collects all KB doc URLs into a set.
+    2. Iterates over all internal WebsiteResource records from Django's DB.
+       Compares each resource's URL against the KB URL set:
+       - "in_kb": resource URL exists in KB — the resource is indexed
+       - "missing_from_kb": resource URL NOT in KB — needs to be added/re-ingested
+    3. Also finds "untracked" docs: URLs present in the KB but not in the
+       internal WebsiteResource table (added to KB outside of this app).
+
+    Returns JSON to the frontend:
+    - resources: list of {id, url, title, status} for each internal resource
+    - untracked: list of {url, title} for KB docs not tracked internally
+    - kb_total: total documents in the KB
+    - internal_total: total WebsiteResource records in Django DB
+    """
+    try:
+        kb_docs = []
+        page = 1
+        while True:
+            data = list_kb_documents(page=page, page_size=50)
+            kb_docs.extend(data.get("documents", []))
+            if len(kb_docs) >= data.get("total", 0):
+                break
+            page += 1
+
+        kb_urls = {doc["url"] for doc in kb_docs if doc.get("url")}
+
+        # compare each internal WebsiteResource urls against KB URLs
+        internal_resources = WebsiteResource.objects.all()
+        results = []
+        internal_urls = set()
+        for resource in internal_resources:
+            internal_urls.add(resource.url)
+            results.append({
+                "id": resource.id,
+                "url": resource.url,
+                "title": resource.title,
+                "status": "in_kb" if resource.url in kb_urls else "missing_from_kb",
+            })
+
+        untracked = [
+            {"url": doc["url"], "title": doc["title"]}
+            for doc in kb_docs
+            if doc.get("url") and doc["url"] not in internal_urls
+        ]
+
+        return JsonResponse({
+            "success": True,
+            "resources": results,
+            "untracked": untracked,
+            "kb_total": len(kb_docs),
+            "internal_total": len(results),
+        })
+    except httpx.ConnectError:
+        return JsonResponse({
+            "success": False,
+            "error": "Could not connect to the Knowledge Base server.",
+        }, status=503)
+    except httpx.HTTPStatusError as e:
+        return JsonResponse({
+            "success": False,
+            "error": f"Knowledge Base server returned an error (HTTP {e.response.status_code}).",
+        }, status=502)
+    except Exception:
+        logger.exception("KB sync failed")
+        return JsonResponse({
+            "success": False,
+            "error": "An unexpected error occurred during sync.",
+        }, status=500)
diff --git a/hospexplorer/hospexplorer/settings.py b/hospexplorer/hospexplorer/settings.py
@@ -161,6 +161,14 @@
 LLM_TIMEOUT = int(os.getenv("LLM_TIMEOUT", 120))
 LLM_MAX_TOKENS = int(os.getenv("LLM_MAX_TOKENS", "4096"))
 
+# MCP Knowledge Base Configuration
+KB_MCP_HOST = os.getenv("KB_MCP_HOST", "http://localhost:8002")
+KB_MCP_JWT_TOKEN = os.getenv("KB_MCP_JWT_TOKEN", "")
+KB_MCP_TIMEOUT = int(os.getenv("KB_MCP_TIMEOUT", 30))
+
+# Number of resources to fetch per page
+KB_RESOURCES_PAGE_SIZE = int(os.getenv("KB_RESOURCES_PAGE_SIZE", 20))
+
 # Sidebar conversations limit
 SIDEBAR_CONVERSATIONS_LIMIT = int(os.getenv("SIDEBAR_CONVERSATIONS_LIMIT", 10))