Memory Persistence Tools

Rigour's MCP server includes tools for context memory persistence - solving the common problem of AI agents losing context and not following user instructions across sessions.

The Problem

AI coding assistants often:

• Forget user preferences between sessions
• Lose track of project-specific conventions
• Repeat mistakes they were told to avoid
• Fail to follow critical instructions

The Solution

Rigour provides three MCP tools for persistent memory:

rigour_remember

Store instructions that persist across sessions:

{
  "name": "rigour_remember",
  "arguments": {
    "cwd": "/path/to/project",
    "key": "coding_style",
    "value": "Always use TypeScript with strict mode. Prefer async/await over callbacks."
  }
}

rigour_recall

Retrieve stored instructions at the start of each session:

{
  "name": "rigour_recall",
  "arguments": {
    "cwd": "/path/to/project"
  }
}

This returns all stored memories. Pass a key to retrieve a specific memory.

rigour_forget

Remove a stored memory:

{
  "name": "rigour_forget",
  "arguments": {
    "cwd": "/path/to/project",
    "key": "deprecated_instruction"
  }
}

Storage Location

Memories are stored in .rigour/memory.json within your project directory. This file is automatically gitignored when you run rigour init.

Best Practices

1. Start sessions with recall: Always call rigour_recall at the start of a new session
2. Use meaningful keys: user_preferences, coding_conventions, critical_warnings
3. Keep values concise: Store actionable instructions, not verbose explanations
4. Clean up: Use rigour_forget to remove outdated instructions

Example Workflow

User: "Remember that I prefer functional programming patterns"

Agent: *calls rigour_remember with key="coding_style"*

--- New session ---

Agent: *calls rigour_recall*
Agent: "I see you prefer functional programming patterns. I'll use that approach."