Graphiti MCP

# Web UI - Example Values to Enter This document provides example values you can use to test the Graphiti MCP Web UI. ## 1. Store Memory Tab ### Example 1: Project Information **Content:** ``` I'm working on a machine learning project to predict customer churn using XGBoost. The project is scheduled for completion in Q2 2024 and uses historical transaction data. ``` **Tags:** ``` project machine-learning churn-prediction in-progress ``` **Metadata:** - Key: `project_name`, Value: `Customer Churn Prediction` - Key: `status`, Value: `in-progress` - Key: `deadline`, Value: `2024-06-30` - Key: `algorithm`, Value: `XGBoost` --- ### Example 2: Technical Details **Content:** ``` The churn prediction model uses features like transaction frequency, average order value, customer lifetime value, and support ticket history. We're achieving 87% accuracy with the current model. ``` **Tags:** ``` project technical model-details ``` **Metadata:** - Key: `project_name`, Value: `Customer Churn Prediction` - Key: `model_accuracy`, Value: `87%` - Key: `features`, Value: `transaction_frequency, avg_order_value, clv, support_tickets` --- ### Example 3: Team Information **Content:** ``` The project team consists of Sarah (data scientist), Mike (ML engineer), and myself. We have weekly standups every Monday at 10 AM. ``` **Tags:** ``` project team meetings ``` **Metadata:** - Key: `project_name`, Value: `Customer Churn Prediction` - Key: `team_size`, Value: `3` - Key: `meeting_schedule`, Value: `Monday 10 AM` --- ### Example 4: Learning Resource **Content:** ``` I completed an online course on deep learning from MIT. Learned about neural networks, backpropagation, and convolutional networks. The course was excellent and I'm planning to apply this knowledge to my next project. ``` **Tags:** ``` learning deep-learning course ``` **Metadata:** - Key: `provider`, Value: `MIT` - Key: `topic`, Value: `deep-learning` - Key: `status`, Value: `completed` --- ### Example 5: Personal Preference **Content:** ``` I prefer Python for data science projects because of its rich ecosystem of libraries like pandas, numpy, and scikit-learn. I also find the syntax more readable than R. ``` **Tags:** ``` preferences programming python data-science ``` **Metadata:** - Key: `category`, Value: `programming_language` - Key: `preference_level`, Value: `high` --- ## 2. Retrieve Memories Tab ### Example Queries: 1. **Simple search:** ``` machine learning projects ``` 2. **Specific topic:** ``` customer churn prediction ``` 3. **Team-related:** ``` team members and meetings ``` 4. **Learning-related:** ``` deep learning courses ``` 5. **Preferences:** ``` programming language preferences ``` 6. **Technical details:** ``` model accuracy and features ``` **Limit:** Start with `10` (default), then try `5` or `20` to see different results. --- ## 3. Get Context Tab ### Example Queries: 1. **Project overview:** ``` Tell me about the customer churn prediction project ``` 2. **Comprehensive summary:** ``` What do you know about my machine learning projects and team? ``` 3. **Learning progress:** ``` What have I learned about deep learning and how am I applying it? ``` 4. **Technical context:** ``` Give me context about the model performance and features used ``` 5. **General inquiry:** ``` What projects am I currently working on? ``` **Max Memories:** Start with `20` (default), increase to `30` or `50` for more comprehensive context. --- ## 4. Create Relationship Tab ### Step 1: Store Two Memories First **Memory 1 (Project Overview):** - Content: `Working on customer churn prediction project using XGBoost` - Store it and **copy the Memory ID** from the result **Memory 2 (Technical Details):** - Content: `Model uses transaction frequency, CLV, and support tickets. Accuracy: 87%` - Store it and **copy the Memory ID** from the result ### Step 2: Create Relationship **Source Memory ID:** (Paste Memory 1 ID here) **Relationship Type:** `has_details` **Target Memory ID:** (Paste Memory 2 ID here) ### Example Relationship Types: - `relates_to` - General relationship - `has_details` - One memory provides details about another - `follows` - Temporal sequence - `references` - One memory references another - `part_of` - One memory is part of a larger concept - `enables` - One memory enables or supports another - `generated` - One memory was generated from another --- ## 5. Search Graph Tab (Cypher Queries) ### Example 1: List All Memories ```cypher MATCH (m:Memory) RETURN m.id as id, m.content as content, m.tags as tags, m.created_at as created_at ORDER BY m.created_at DESC LIMIT 20 ``` ### Example 2: Find Memories by Tag ```cypher MATCH (m:Memory) WHERE 'project' IN m.tags RETURN m.id as id, m.content as content, m.tags as tags ORDER BY m.created_at DESC ``` ### Example 3: Find Memories with Relationships ```cypher MATCH (m:Memory)-[r]->(n:Memory) RETURN m.id as source_id, m.content as source_content, type(r) as relationship, n.id as target_id, n.content as target_content LIMIT 20 ``` ### Example 4: Count Memories by Tag ```cypher MATCH (m:Memory) UNWIND m.tags as tag RETURN tag, count(*) as count ORDER BY count DESC ``` ### Example 5: Find Most Recent Memories ```cypher MATCH (m:Memory) RETURN m.id as id, m.content as content, m.created_at as created_at ORDER BY m.created_at DESC LIMIT 10 ``` ### Example 6: Search by Metadata (contains) ```cypher MATCH (m:Memory) WHERE m.metadata CONTAINS 'Customer Churn Prediction' RETURN m.id as id, m.content as content, m.metadata as metadata ``` ### Example 7: Find Connected Memories (2 hops) ```cypher MATCH path = (start:Memory)-[*1..2]-(related:Memory) RETURN DISTINCT related.id as id, related.content as content, length(path) as distance ORDER BY distance LIMIT 20 ``` --- ## 6. Browse All Tab Simply click the **"Load All Memories"** button. No input needed! This will show all memories stored in the database (up to 50 by default). --- ## Quick Test Workflow 1. **Store 3-4 memories** using examples from section 1 2. **Copy the Memory IDs** from the results 3. **Create relationships** between related memories (section 4) 4. **Retrieve memories** using queries from section 2 5. **Get context** using queries from section 3 6. **Browse all** to see everything you've stored 7. **Search graph** with Cypher queries from section 5 --- ## Tips - **Start simple**: Store a few memories first before trying relationships - **Use descriptive tags**: Makes retrieval easier - **Add metadata**: Helps with organization and filtering - **Create relationships**: Builds a knowledge graph structure - **Experiment with queries**: Try different search terms to see what works best - **Use Cypher for advanced queries**: When simple search isn't enough --- ## Troubleshooting **No memories found?** - Make sure you've stored at least one memory first - Try a broader search query - Check the "Browse All" tab to see what's actually stored **Can't create relationship?** - Verify both Memory IDs are correct (copy-paste from store results) - Make sure both memories exist (check Browse All tab) - Memory IDs are UUIDs - they're long strings **Cypher query errors?** - Check syntax (MATCH, RETURN, WHERE clauses) - Make sure property names match (tags, content, metadata) - Start with simple queries and build up complexity