1. Open the GPT Store
Go to: https://chatgpt.com/gpts You will see a page like this:
2. Basic GPT Setup
See the configuration page reference below:
Name
Copy
ChatGPT with MemMachine
Description
Copy
I am an intelligent assistant with long-term memory powered by MemMachine.
Instructions
Copy the following:Copy
You MUST actively use memory on EVERY message to provide personalized responses.
## CRITICAL: Session Values - MUST USE EXACTLY AS SHOWN
When calling SearchMemory or AddMemory, you MUST ALWAYS use these EXACT session values:
{
"group_id": "YourName",
"user_id": ["YourName"],
"session_id": "YourName",
"agent_id": ["YourName"]
}
Also use these exact values:
- producer: "YourName"
- produced_for: "YourName"
## CRITICAL: Memory Operations on EVERY Message
On EVERY user message, you MUST:
1. ALWAYS start by searching memories (SearchMemory):
- Search for relevant past information based on the user's new message
- Use the results to understand context, preferences, history, long-term facts
2. ALWAYS respond naturally using the retrieved memories:
- Do NOT mention "searching memory"
- Use context naturally in your reply
3. ALWAYS store new information (AddMemory):
- Store any new facts, preferences, goals, or plans the user shares
"YourName" with any test user ID you want.
3. Enable Capabilities
Scroll to Capabilities and enable:- Code Interpreter & Data Analysis
4. Add MemMachine Action
Click: Create new action You will see this page:
OpenAPI Schema (MemMachine)
Copy
{
"openapi": "3.1.0",
"info": {
"title": "MemMachine Memory API",
"description": "API for storing and retrieving long-term memories. Use this to remember important information about users across conversations. Automatically stores memories in both episodic memory (conversation history) and profile memory (long-term user traits and facts).",
"version": "v1.0.0"
},
"servers": [
{
"url": "A HTTPS URL hosting MemMachine",
"description": "MemMachine API server"
}
],
"paths": {
"/v1/memories": {
"post": {
"description": "CRITICAL: Use this on EVERY message to store new information. Automatically store any new facts, preferences, plans, or details the user shares. Include full context in episode_content. This ensures future conversations have complete context.",
"summary": "Store a new memory",
"operationId": "AddMemory",
"x-openai-isConsequential": false,
"requestBody": {
"required": true,
"content": {
"application/json": {
"schema": {
"$ref": "#/components/schemas/MemoryInput"
}
}
}
},
"responses": {
"200": {
"description": "Memory successfully stored",
"content": {
"application/json": {
"schema": {
"type": "object",
"properties": {
"status": {
"type": "integer",
"description": "Status code (0 for success)",
"example": 0
}
}
}
}
}
},
"400": {
"description": "Bad request - invalid producer or produced_for ID",
"content": {
"application/json": {
"schema": {
"$ref": "#/components/schemas/ErrorResponse"
}
}
}
},
"404": {
"description": "No matching episodic memory instance found",
"content": {
"application/json": {
"schema": {
"$ref": "#/components/schemas/ErrorResponse"
}
}
}
},
"500": {
"description": "Internal server error",
"content": {
"application/json": {
"schema": {
"$ref": "#/components/schemas/ErrorResponse"
}
}
}
}
},
"deprecated": false
}
},
"/v1/memories/search": {
"post": {
"description": "CRITICAL: Use this on EVERY message to retrieve context. Always search for relevant past information before responding. This searches both profile memory (long-term user traits) and episodic memory (past conversations) to provide personalized responses.",
"summary": "Search memories",
"operationId": "SearchMemory",
"x-openai-isConsequential": false,
"requestBody": {
"required": true,
"content": {
"application/json": {
"schema": {
"$ref": "#/components/schemas/SearchInput"
}
}
}
},
"responses": {
"200": {
"description": "Search results returned successfully",
"content": {
"application/json": {
"schema": {
"$ref": "#/components/schemas/SearchResult"
}
}
}
},
"400": {
"description": "Bad request",
"content": {
"application/json": {
"schema": {
"$ref": "#/components/schemas/ErrorResponse"
}
}
}
},
"404": {
"description": "No matching episodic memory instance found",
"content": {
"application/json": {
"schema": {
"$ref": "#/components/schemas/ErrorResponse"
}
}
}
},
"500": {
"description": "Internal server error",
"content": {
"application/json": {
"schema": {
"$ref": "#/components/schemas/ErrorResponse"
}
}
}
}
},
"deprecated": false
}
}
},
"components": {
"schemas": {
"MemoryInput": {
"type": "object",
"description": "Input for storing a new memory",
"properties": {
"session": {
"$ref": "#/components/schemas/SessionData",
"description": "Session context. Fixed values are required."
},
"producer": {
"type": "string",
"default": "patient_001",
"description": "The ID of the entity that produced the content (usually the user ID). Must be 'patient_001'.",
"example": "patient_001"
},
"produced_for": {
"type": "string",
"default": "patient_001",
"description": "The ID of the entity the content was produced for (usually 'patient_001' or the agent ID). Must be 'patient_001'.",
"example": "patient_001"
},
"episode_content": {
"type": "string",
"description": "The memory content to store. Include full context - not just a snippet. Always prepend with current date in YYYY-MM-DD format if you know the user's timezone, UTC otherwise.",
"example": "2024-11-04: User prefers Python over JavaScript for backend development"
},
"episode_type": {
"type": "string",
"description": "Type of memory: 'message' (conversation), 'preference' (user preference), 'fact' (factual information), 'goal' (user goal), etc.",
"default": "message",
"example": "preference"
},
"metadata": {
"type": "object",
"description": "Optional metadata",
"additionalProperties": true,
"default": {}
}
},
"required": [
"session",
"producer",
"produced_for",
"episode_content",
"episode_type"
],
"additionalProperties": true
},
"SearchInput": {
"type": "object",
"description": "Input for searching memories",
"properties": {
"session": {
"$ref": "#/components/schemas/SessionData",
"description": "Session context. Fixed values are required."
},
"query": {
"type": "string",
"description": "Natural language query to search for relevant memories",
"example": "user preferences about programming languages"
},
"limit": {
"type": "integer",
"description": "Maximum number of results to return. Default is 5.",
"default": 5,
"minimum": 1,
"maximum": 50
},
"filter": {
"type": "object",
"description": "Optional filters for episodic/profile memory",
"additionalProperties": true,
"nullable": true
}
},
"required": [
"session",
"query"
],
"additionalProperties": true
},
"SessionData": {
"type": "object",
"description": "Session context (fixed to a single identity). MUST use: group_id='patient_001', user_id=['patient_001'], session_id='patient_001', agent_id=['patient_001'].",
"properties": {
"group_id": {
"type": "string",
"const": "patient_001",
"default": "patient_001",
"description": "Fixed group ID."
},
"user_id": {
"type": "array",
"minItems": 1,
"maxItems": 1,
"default": [
"patient_001"
],
"items": {
"type": "string",
"const": "patient_001"
},
"description": "Fixed user list with a single value."
},
"session_id": {
"type": "string",
"const": "patient_001",
"default": "patient_001",
"description": "Fixed session ID."
},
"agent_id": {
"type": "array",
"minItems": 1,
"maxItems": 1,
"default": [
"patient_001"
],
"items": {
"type": "string",
"const": "patient_001"
},
"description": "Fixed agent list with a single value."
}
},
"required": [
"group_id",
"user_id",
"session_id",
"agent_id"
],
"additionalProperties": false
},
"SearchResult": {
"type": "object",
"description": "Search results containing memories from both episodic and profile memory",
"properties": {
"status": {
"type": "integer",
"description": "Status code (0 for success)",
"example": 0
},
"content": {
"type": "object",
"description": "Search results containing episodic_memory and profile_memory arrays",
"properties": {
"episodic_memory": {
"type": "array",
"description": "Results from episodic memory (conversation history)",
"items": {
"type": "object",
"additionalProperties": true
}
},
"profile_memory": {
"type": "array",
"description": "Results from profile memory (long-term user traits and facts)",
"items": {
"type": "object",
"additionalProperties": true
}
}
}
}
},
"required": [
"status",
"content"
]
},
"ErrorResponse": {
"type": "object",
"description": "Error response",
"properties": {
"detail": {
"type": "string",
"description": "Error message"
}
}
}
},
"securitySchemes": {}
},
"security": []
}
Important
In the schema, update:Copy
"servers": [
{
"url": "A HTTPS URL hosting MemMachine",
"description": "MemMachine API server"
}
]
"A HTTPS URL hosting MemMachine" with your actual HTTPS MemMachine endpoint URL.
5. Test the Actions
Once the schema is saved, you will see two Available Actions:- AddMemory
- SearchMemory
- Run a simple SearchMemory query
- Run a simple AddMemory request