Auto-Save Important Decisions in Cursor

This scenario shows how to use Cursor with YoMemo MCP so the AI proactively and securely saves important decisions and preferences—tech stack, business rules, coding habits—with end-to-end encryption. In new chats or when starting a new feature, the AI loads relevant memories automatically so you don’t have to repeat yourself.

Why Auto-Save Important Decisions

• Less repetition: Tech stack, business rules, and preferences are stated once; later conversations use memories.
• Secure and traceable: All memories are encrypted on the client before upload; the server cannot decrypt them; only you hold the key.
• Low friction: No need to say “remember this” or “save this”—the AI calls save_memory when it detects important information, and replies with ✓ after saving.

Prerequisites

• YoMemo MCP (yomemoai-mcp) is configured in Cursor with your API key and RSA private key path.
• save_memory and load_memories work correctly.

Configure the Proactive Memory Rule

Add a rule in Cursor that tells the AI when to save and when to load memories. Two options:

Option 1: Cursor global / project Rules for AI

1. Open Cursor settings → Rules for AI.
2. Add a new rule and paste the following:

You are equipped with Yomemo.ai MCP.

## When to use `save_memory`:
- **Tech Stack**: When we decide on a specific library or version.
- **Business Logic**: When I explain a complex internal rule.
- **Preferences**: If I tell you "I prefer using early returns in Go".

## When to use `load_memories`:
- At the start of a new feature implementation, check if there's relevant context in the 'coding' or 'project-name' handle.

## Feedback:
- After saving, just add a ✓ in your response. No need for a long confirmation.

3. Save. In that project or globally, the AI will follow this rule and call save_memory / load_memories proactively.

Tip

Replace project-name in the rule with your project name or a handle you use (e.g. memo, yomemoai) so memories are organized per project.

Option 2: Project-level rule file (.cursor/rules)

To apply the rule only in the current project:

1. Create .cursor/rules/ in the project root (if it doesn’t exist).
2. Add a file there, e.g. yomemo-memory.mdc.
3. Paste the full rule from the block above into that file and save.

The AI will then use this proactive memory rule only when Cursor is opened in this project.

Rule Summary

Scenario	Behavior
Tech Stack	When you agree on a library, framework, or version (e.g. “we’ll use React 19”, “Go 1.24”), the AI should call save_memory with that decision, e.g. under handle coding or the project name.
Business / internal rules	When you explain a complex internal rule (e.g. “orders unpaid after 3 days are auto-cancelled”), the AI should save it for later implementation or debugging.
Preferences	When you state a coding or tool preference (e.g. “I prefer early returns in Go”, “use Tailwind, no inline styles”), the AI should save it so future replies follow it.
Start of new feature / new chat	Before implementing, the AI should call load_memories and check handles like coding or the project handle for relevant tech stack, rules, and preferences.
After saving	After a successful save, the AI adds ✓ in its reply—no long confirmation, so the conversation stays smooth.

What You Get

Proactive save

The AI calls save_memory when it recognizes preferences, decisions, or reusable logic—no need to say “remember this” every time.

Low-friction feedback

After saving, the AI just adds ✓ so the flow isn’t interrupted.

Auto-load in new chats

When starting a new chat or feature, the AI calls load_memories and uses that context in its answers.

End-to-end encryption

All memories are encrypted locally before upload; the server cannot decrypt them. See Security.

How This Relates to YoMemo MCP

This scenario depends on the Python MCP integration (yomemoai-mcp). The rule only defines when the AI calls save_memory / load_memories; encryption, storage, and retrieval are done by the MCP and YoMemo backend. If MCP isn’t set up yet, follow the Getting Started MCP configuration first.

For more details, see CURSOR_RULES.md in the python-yomemo-mcp repo.

Next Steps

• Adjust the handle in the rule (e.g. coding, project name) so memories are grouped by project or purpose.
• Learn about save_memory’s idempotent_key for updating the same decision without creating duplicate memories.
• Explore other use cases and integrations.