This directory contains a toolset package that enables integration of MemMachine memory operations into FastGPT workflows.
MemMachine provides a toolset that integrates with FastGPT workflows, enabling AI agents to have persistent memory capabilities. This allows agents to retain past interactions, user preferences, and context across multiple sessions.
- Install FastGPT by following the official documentation.
- Log in to the FastGPT GUI using the root user.
- Go to the admin page.
- Download the MemMachine toolset package
- Click the "Add resources" button and select "Import/update resources".
- Choose the MemMachine toolset package to upload.
- After verifying the file information, click "Confirm import".
Store important information about the user or conversation into memory.
Parameters
| Parameter | Type | Default | Description |
|---|---|---|---|
| orgId | string | universal | Unique identifier for the organization |
| projectId | string | universal | Unique identifier for the project |
| types | multipleSelect | ['episodic', 'semantic'] | Memory types; leave empty to add to all types |
| content | string | - | Required Memory content to store |
| producer | string | user | Sender of the memory content |
| producedFor | string | - | Recipient of the memory content |
| timestamp | string | - | Creation time of the memory content (ISO 8601 format) |
| role | string | - | Role of the memory content in the conversation |
| metadata | string | - | Additional memory attributes (in JSON format) |
Output
{
"memoryId": "New memory ID"
}Retrieve relevant context, memories, or profile for a user.
Parameters
| Parameter | Type | Default | Description |
|---|---|---|---|
| orgId | string | universal | Unique identifier for the organization |
| projectId | string | universal | Unique identifier for the project |
| types | multipleSelect | ['episodic', 'semantic'] | Memory types; leave empty to search all types |
| query | string | - | Required Natural language query for memory retrieval |
| limit | number | 10 | Required Maximum number of search results |
| filter | string | - | Condition to filter memories |
| contextTemplate | string | default template | Template for building memory context |
Output
{
"memoryContext": "memory context"
}