adds withSupermemory wrapper and input/output processors for
mastra agents:
- input processor fetches and injects memories into system prompt
before llm calls
- output processor saves conversations to supermemory after
responses
- supports profile, query, and full memory search modes
- includes custom prompt templates and requestcontext support
const agent = new Agent(withSupermemory(
{ id: "my-assistant", model: openai("gpt-4o"), instructions:
"..." },
"user-123",
{ mode: "full", addMemory: "always", threadId: "conv-456" }
))
includes docs as well
this pr also reworks how the tools package works into shared modules
## Add customizable prompt templates for memory injection
**Changes:**
- Add `promptTemplate` option to `withSupermemory()` for full control over injected memory format (XML, custom branding, etc.)
- New `MemoryPromptData` interface with `userMemories` and `generalSearchMemories` fields
- Exclude `system` messages from persistence to avoid storing injected prompts
- Add JSDoc comments to all public interfaces for better DevEx
**Usage:**
```typescript
const customPrompt = (data: MemoryPromptData) => `
<user_memories>
${data.userMemories}
${data.generalSearchMemories}
</user_memories>
`.trim()
const model = withSupermemory(openai("gpt-4"), "user-123", {
promptTemplate: customPrompt,
})
```
### Added streaming support to the Supermemory middleware and improved memory handling in the AI SDK integration.
### What changed?
- Refactored the middleware architecture to support both streaming and non-streaming responses
- Extracted memory prompt functionality into a separate module (`memory-prompt.ts`)
- Added memory saving capability for streaming responses
- Improved the formatting of memory content with a "User Supermemories:" prefix
- Added utility function to filter out supermemories from content
- Created a new streaming example in the test app with a dedicated route and page
- Updated version from 1.3.0 to 1.3.1 in package.json
- Simplified installation instructions in [README.m](http://README.md)d
### TL;DR
Added OpenAI SDK middleware support for SuperMemory integration, allowing direct memory injection without AI SDK dependency.
### What changed?
- Added `withSupermemory` middleware for OpenAI SDK that automatically injects relevant memories into chat completions
- Implemented memory search and injection functionality for OpenAI clients
- Restructured the OpenAI module to separate tools and middleware functionality
- Updated README with comprehensive documentation and examples for the new OpenAI middleware
- Added test implementation with a Next.js API route example
- Reorganized package exports to support the new structure
### TL;DR
Added support for conversation grouping in Supermemory middleware through a new `conversationId` parameter.
### What changed?
- Added a new `conversationId` option to the `withSupermemory` function to group messages into a single document for contextual memory generation
- Updated the middleware to use this conversation ID when adding memories, using a `customId` format of `conversation:{conversationId}`
- Created a new `getConversationContent` function that extracts the full conversation content from the prompt parameters
- Enhanced memory storage to save entire conversations rather than just the last user message
- Updated documentation and examples to demonstrate the new parameter usage
### How to test?
1. Import the `withSupermemory` function from the package
2. Create a model with memory using the new `conversationId` parameter:
```typescript
const modelWithMemory = withSupermemory(openai("gpt-4"), "user-123", {
conversationId: "conversation-456",
mode: "full",
addMemory: "always"
})
```
3. Use the model in a conversation and verify that messages are grouped by the conversation ID
4. Check that memories are being stored with the custom ID format `conversation:{conversationId}`
### Why make this change?
This enhancement improves the contextual understanding of the AI by allowing related messages to be grouped together as a single conversation document. By using a conversation ID, the system can maintain coherent memory across multiple interactions within the same conversation thread, providing better context retrieval and more relevant responses.
### TL;DR
Added support for automatically saving user messages to Supermemory.
### What changed?
- Added a new `addMemory` option to `wrapVercelLanguageModel` that accepts either "always" or "never" (defaults to "never")
- Implemented the `addMemoryTool` function to save user messages to Supermemory
- Modified the middleware to check the `addMemory` setting and save the last user message when appropriate
- Initialized the Supermemory client in the middleware to enable memory storage
### How to test?
1. Set the `SUPERMEMORY_API_KEY` environment variable
2. Use the `wrapVercelLanguageModel` function with the new `addMemory: "always"` option
3. Send a user message through the model
4. Verify that the message is saved to Supermemory with the specified container tag
### Why make this change?
This change enables automatic memory creation from user messages, which improves the system's ability to build a knowledge base without requiring explicit memory creation calls. This is particularly useful for applications that want to automatically capture and store user interactions for future reference.
