Knowledge Graph Memory Server
Build and manage a persistent knowledge graph by creating, connecting, and retrieving memories.
Example Prompts
Create a project memory called "New Website Launch" and connect task memories
like "Design Homepage," "Setup Analytics," and "Launch Marketing Campaign."
Knowledge Graph for Developer Workflows
The Memory integration creates a connected knowledge graph that persists across your development workflow. This enables your team to build a comprehensive understanding of your project architecture, codebase structure, and development progress.
In the diagram above, you can see how different memory types work together to form a complete picture of your development ecosystem:
-
Project Memory: Acts as the central hub that connects all aspects of your project
-
Code Structure Nodes: Track your codebase architecture
- Components: UI elements, their behaviors, and rendering logic
- Data Models: Your data schemas, types, and state management
- API Layer: Endpoints, services, and external integrations
-
Development Tasks: Manage your team's workflow
- Feature Tasks: Track new feature development (like authentication)
- Bug Memories: Document issues, their root causes, and solutions
- Todo Items: Track smaller tasks that need completion
-
Team Knowledge: Preserve important decisions and discussions
- Decision Memories: Document key architectural choices
- Meeting Notes: Capture important discussions and their outcomes
The power of this approach comes from the relationships between nodes. For example:
- A Feature Task may DEPEND_ON certain APIs to be completed
- A Bug may be RELATED_TO specific data models
- A Decision may INFLUENCE how components are structured
- A Todo might be PART_OF a larger feature implementation
By querying these relationships, your AI can provide contextual assistance based on the full development history, helping new team members onboard faster and ensuring institutional knowledge isn't lost.
Overview
The Memory integration enables your AI to create and manage a persistent knowledge graph. Each memory can store different types of information and be connected to other memories through various relationship types, allowing for complex knowledge structures and contextual recall.
Memory is built into Skeet and requires no additional authentication.
When to Use
Memory Integration is ideal for:
- Building persistent knowledge bases
- Managing project-related information
- Tracking conversations and their context
- Organizing tasks and their dependencies
- Managing configuration settings
- Tracking financial information
- Creating and managing todo lists
- Building connected knowledge graphs
Create a new project memory for our website redesign initiative
How It Works
The Memory integration provides several tools to manage your knowledge graph:
- Create memories of different types
- Retrieve specific memories by ID
- Search memories by type or content
- Update existing memories
- Create relationships between memories
- Query related memories
- Delete memories when no longer needed
1. Create a project memory
2. Add related task memories
3. Connect tasks with dependencies
4. Update task status as work progresses
5. Query related tasks to track progress
Memory Types
The following memory types are supported:
- Conversation: Store dialogue and discussion context
- Topic: Organize information around specific subjects
- Project: Track project-related information
- Task: Manage individual tasks and their status
- Issue: Track problems and their resolutions
- Config: Store configuration settings
- Finance: Track financial information
- Todo: Manage todo items
Parameters
The Memory integration tools accept various parameters depending on the operation:
- Name
type
- Type
- string
- Description
The type of memory (Conversation, Topic, Project, Task, Issue, Config, Finance, Todo).
- Name
content
- Type
- string
- Description
The main content of the memory.
- Name
title
- Type
- string
- Description
Optional title for the memory.
- Name
metadata
- Type
- object
- Description
Additional metadata for the memory.
- Name
id
- Type
- string
- Description
Identifier for retrieving or updating specific memories.
- Name
status
- Type
- string
- Description
Status information for tasks or issues.
- Name
dueDate
- Type
- number
- Description
Timestamp for due dates on tasks or todos.
Examples
Here are examples of how to use Memory integration for different scenarios:
Create a new project memory for our AI integration initiative
The AI will use the memory tools to manage these operations:
{
"type": "Project",
"title": "AI Integration Initiative",
"content": "Project to integrate AI capabilities into our platform",
"metadata": {
"startDate": "2024-03-20",
"priority": "high"
}
}
// Create related task
{
"type": "Task",
"title": "Setup API Endpoints",
"content": "Create necessary API endpoints for AI integration",
"status": "pending",
"dueDate": 1711238400000
}
// Create relationship
{
"fromId": "project-123",
"toId": "task-456",
"type": "CONTAINS",
"properties": {
"order": 1
}
}
Advanced Usage
Memory integration supports advanced features for complex knowledge management:
Relationship Types
You can create different types of relationships between memories:
- CONTAINS: Hierarchical relationship
- RELATED_TO: General association
- DEPENDS_ON: Dependency relationship
- PART_OF: Component relationship
- RESOLVED_BY: Issue resolution relationship
- CREATED_AT: Temporal relationship
- UPDATED_AT: Update relationship
{
"fromId": "task-123",
"toId": "task-456",
"type": "DEPENDS_ON",
"properties": {
"criticality": "high"
}
}
Search and Query
You can search memories by type and content, and query related memories:
Search for all high-priority tasks in the AI integration project
The AI might process this as:
{
"type": "Task",
"keyword": "high priority AI integration",
"topResults": 5
}
What's next?
Now that you understand Memory integration, explore other integrations to make your AI even more powerful.
Integrations
Sequential Thinking
Enable dynamic and reflective problem-solving through structured thinking steps.