# Qdrant Point Details Resource
## Overview
The `qdrant://collection/{collection_name}/{point_id}` resource provides detailed information about a specific point stored in a Qdrant collection, with automatic decompression and structured output.
## URI Pattern
```
qdrant://collection/{collection_name}/{point_id}
```
## Parameters
- `collection_name`: Name of the Qdrant collection (e.g., `mcp_tool_responses`)
- `point_id`: Unique identifier of the point (UUID format, e.g., `88fc2b49-0e61-4617-a9cb-02812375394a`)
## Response Structure (QdrantPointDetailsResponse)
```json
{
"qdrant_enabled": true,
"collection_name": "mcp_tool_responses",
"point_id": "88fc2b49-0e61-4617-a9cb-02812375394a",
"point_exists": true,
"payload": {
// Full raw payload from Qdrant
},
"vector_available": false,
"vector_size": null,
// Extracted common fields for easy access:
"tool_name": "search_gmail_messages",
"user_email": "user@example.com",
"timestamp": "2025-01-15T10:30:00Z",
"session_id": "abc123",
"payload_type": "tool_response",
"compressed": true,
// Decompressed data:
"response_data": {
// Parsed JSON response from the tool
},
"retrieved_at": "2025-01-15T11:00:00Z"
}
```
## Usage Examples
### 1. Basic Access via MCP Client
```
Access resource: qdrant://collection/mcp_tool_responses/88fc2b49-0e61-4617-a9cb-02812375394a
```
### 2. Template Usage (Coming Soon)
Once templates are implemented, you'll be able to use this resource in Jinja2 templates:
```jinja2
{# Get point details #}
{% set point = resource('qdrant://collection/mcp_tool_responses/88fc2b49-0e61-4617-a9cb-02812375394a') %}
{# Access structured fields #}
<div class="point-details">
<h3>{{ point.tool_name }}</h3>
<p>User: {{ point.user_email }}</p>
<p>Timestamp: {{ point.timestamp }}</p>
{# Access decompressed response data #}
{% if point.response_data %}
<pre>{{ point.response_data | tojson(indent=2) }}</pre>
{% endif %}
</div>
```
### 3. Dynamic Point ID from Search
```jinja2
{# First search for relevant points #}
{% set search_results = resource('qdrant://search/gmail messages from john') %}
{# Then get details for the first result #}
{% if search_results.results %}
{% set point_id = search_results.results[0].id %}
{% set details = resource('qdrant://collection/mcp_tool_responses/' + point_id) %}
<div>
<h4>{{ details.tool_name }}</h4>
<p>{{ details.response_data }}</p>
</div>
{% endif %}
```
### 4. Build Lists from Cache
```jinja2
{# Get cache organized by tool #}
{% set cache = resource('qdrant://cache') %}
{# Display recent points from a specific tool #}
{% if 'search_gmail_messages' in cache.cache_data %}
<h3>Recent Gmail Searches</h3>
<ul>
{% for point_meta in cache.cache_data.search_gmail_messages[:5] %}
{% set point = resource('qdrant://collection/mcp_tool_responses/' + point_meta.point_id) %}
<li>
<strong>{{ point.timestamp }}</strong> - {{ point.user_email }}
<br>Response: {{ point.response_data.total_results }} messages found
</li>
{% endfor %}
</ul>
{% endif %}
```
## Features
### Automatic Data Decompression
- Compressed payloads are automatically decompressed
- JSON data is automatically parsed
- Errors are handled gracefully with detailed error messages
### Structured Access
- Common metadata fields are extracted to top-level for easy access
- Raw payload is still available for advanced use cases
- Response data is presented in parsed, usable format
### Error Handling
- Returns `point_exists: false` if point not found
- Provides detailed error messages
- Validates collection existence
## Architecture
The resource follows the established Qdrant middleware pattern:
1. **Client Request** → `qdrant://collection/X/Y`
2. **Middleware Intercepts** → [`QdrantUnifiedMiddleware.on_read_resource`](../middleware/qdrant_middleware.py:295)
3. **Handler Processes** → [`QdrantResourceHandler._handle_point_details`](../middleware/qdrant_core/resource_handler.py)
4. **Returns Pydantic Model** → [`QdrantPointDetailsResponse`](../middleware/qdrant_types.py)
5. **Middleware Converts** → `model.model_dump()` for MCP
6. **Client Receives** → Structured JSON response
## Related Resources
- `qdrant://collections/list` - List all collections
- `qdrant://collection/{name}/info` - Collection metadata
- `qdrant://collection/{name}/responses/recent` - Recent points in collection
- `qdrant://search/{query}` - Semantic search to find point IDs
- `qdrant://cache` - Get all point IDs organized by tool
## Use Cases
1. **Detailed Inspection** - Examine specific tool responses in detail
2. **Search → Details Pattern** - Search for points, then get full details
3. **Cache → Details Pattern** - List cache, then fetch specific points
4. **Template Rendering** - Build rich displays of historical tool responses
5. **Debugging** - Inspect exact payloads and metadata for troubleshooting
6. **Analytics** - Access response data for analysis and reporting
## Implementation Files
- **Type Definition**: [`middleware/qdrant_types.py`](../middleware/qdrant_types.py) - `QdrantPointDetailsResponse`
- **Handler Logic**: [`middleware/qdrant_core/resource_handler.py`](../middleware/qdrant_core/resource_handler.py) - `_handle_point_details()`
- **Resource Registration**: [`middleware/qdrant_core/resources.py`](../middleware/qdrant_core/resources.py) - `get_qdrant_point_details()`
- **Middleware Hook**: [`middleware/qdrant_middleware.py`](../middleware/qdrant_middleware.py) - `on_read_resource()`
