Skip to content

← All guides

Archive memory guide

This Continue guide is preserved for existing users. Memory is now supporting infrastructure under the production-action control product.

How to Add Persistent Memory to Continue

Continue is the leading open-source AI code assistant for VS Code and JetBrains, but it does not persist context across sessions. The Novyx MCP Server gives Continue persistent memory — your assistant remembers your codebase, conventions, and decisions across every session and every IDE.

The Problem

  • Continue resets context when you start a new chat or close the IDE
  • Context providers give file access but not persistent memory
  • No semantic search across past conversations or decisions
  • No rollback if the assistant internalizes incorrect project assumptions

The Solution: Novyx MCP Server

The Novyx MCP Server exposes memory tools to Continue via the Model Context Protocol. Continue can store architecture decisions, recall past context, preserve audit context, and recover supported state — all persisted across sessions.

Setup

1. Install the MCP server:

bash
pip install novyx-mcp

2. Add to your Continue config (~/.continue/config.json):

json
{
  "experimental": {
    "modelContextProtocolServers": [
      {
        "transport": {
          "type": "stdio",
          "command": "python",
          "args": ["-m", "novyx_mcp"],
          "env": { "NOVYX_API_KEY": "YOUR_API_KEY" }
        }
      }
    ]
  }
}

This works in both VS Code and JetBrains. Continue auto-discovers the MCP tools and makes them available in your chat sessions.

What Continue Can Do With Memory

Remember across IDEs

"This project uses a monorepo with Turborepo." — stored once in VS Code, recalled instantly in JetBrains.

Recall past debugging sessions

"What was the fix for that CORS issue?" — semantic search finds it even if you use different words.

Track team conventions

"All API routes use zod validation and return typed responses." — the whole team shares the same memory.

Recover wrong context

"Go back to the last good project-memory checkpoint." — supported recovery with audit context.

MCP Tools Available

CategoryTools
Core Memoryremember, recall, forget, list_memories, memory_stats, rollback, audit, link_memories, add_triple, query_triples
Context Spacescreate_space, list_spaces, space_memories, update_space, delete_space, share_space
Replayreplay_timeline, replay_snapshot, replay_lifecycle, replay_diff
Experimental maintenanceOptional memory-maintenance helpers where enabled

See the full MCP server docs on the Integrations page.

Frequently Asked Questions

How do I add persistent memory to Continue?

Install novyx-mcp and add it to your Continue MCP configuration in config.json. Continue gains MCP memory tools so coding sessions can persist across restarts with semantic search and supported recovery workflows.

Does Continue support MCP servers?

Yes. Continue has native MCP support. You configure MCP servers in your Continue config.json file under the experimental.modelContextProtocolServers section.

Can I use Novyx memory across Continue and other editors?

Yes. Novyx is a standalone memory service. Memories stored in Continue are accessible from Cursor, Claude Code, Windsurf, or any tool that connects to the same Novyx API key. Your project knowledge follows you across editors.

Start with the production-action gate

For the current product path, start with a governed production action.

Run the gate