ogham-mcp

Overview

Ogham MCP provides a persistent memory layer for AI coding agents that works across different clients like Claude Code, OpenCode, and Cursor. It solves the problem of AI agents forgetting context between sessions by implementing a searchable shared memory with 97.2% retrieval accuracy. The system supports hybrid search combining semantic and keyword matching, multiple profiles for different contexts, and can export to Obsidian format. It's built with Python and uses PostgreSQL with vector embeddings for storage and retrieval.

Try asking AI

After installing, here are 5 things you can ask your AI assistant:

• memory

  Store a fact or memory in the shared memory system

  Try:Remember that we use PostgreSQL for our databaseTry:Note that our API rate limit is 1000 requests per minute
• search

  Search for memories in the shared memory using hybrid semantic and keyword search

  Try:Find information about our authentication systemTry:Search for previous discussions about database optimization
• list

  List recent memories from the shared memory system

  Try:Show me the last 10 memoriesTry:List all recent decisions about the API design
• delete

  Delete a specific memory by its ID

  Try:Remove memory with ID 12345Try:Delete the memory about the deprecated endpoint
• use_profile

  Switch to a specific memory profile

  Try:Switch to the work profileTry:Use the project-alpha profile for this session
• profiles

  List available memory profiles and their memory counts

  Try:Show all profilesTry:List how many memories are in each profile
• stats

  Show profile statistics and memory usage information

  Try:Show memory statisticsTry:Display profile metrics
• export

  Export memories to a file

  Try:Export all memories to backup.jsonTry:Create a backup of my work profile memories
• import

  Import memories from a file

  Try:Import memories from backup.jsonTry:Load memories from the project file
• cleanup

  Remove expired memories from the shared memory system

  Try:Clean up expired memoriesTry:Remove old memories that are past their retention date
• hybrid_search

  Search across multiple memory profiles simultaneously

  Try:Search for architecture decisions across work and shared profilesTry:Find information about the API in both project-alpha and work profiles
• hooks_recall

  Load project context from memory hooks

  Try:Recall previous project contextTry:Load my work session from memory

Comparable tools

Installation

uvx --from ogham-mcp ogham init

export SUPABASE_URL=https://your-project.supabase.co
export SUPABASE_KEY=your-service-role-key
export EMBEDDING_PROVIDER=openai
export OPENAI_API_KEY=sk-...

claude mcp add ogham -- uvx ogham-mcp

{
  "mcp": {
    "ogham": {
      "type": "local",
      "command": ["uvx", "ogham-mcp"],
      "environment": {
        "SUPABASE_URL": "https://your-project.supabase.co",
        "SUPABASE_KEY": "{env:SUPABASE_KEY}",
        "EMBEDDING_PROVIDER": "openai",
        "OPENAI_API_KEY": "{env:OPENAI_API_KEY}"
      }
    }
  }
}

FAQ

What databases does Ogham support?

Ogham supports Supabase (default), Neon, and self-hosted PostgreSQL databases.

How does Ogham ensure data privacy?

Ogham stores data in your own database. You control the connection credentials and data never leaves your infrastructure unless you use external embedding providers like OpenAI.

Compare ogham-mcp with