Claude Code Integration

Claude Code can Use RetainDB through MCP for:

• grounded project retrieval
• local code exploration
• durable memory for preferences and decisions
• context handoff across sessions

Setup

Follow Claude Code MCP Setup Guide to install the server and configure RETAINDB_API_KEY, RETAINDB_API_URL, and an optional default project.

Primary verbs to use in Claude Code

Start with these:

• search
• search_code
• grep
• read
• explore
• research
• remember
• record
• learn
• share_context

Use namespaced tools like context.query, memory.add, or code.search_semantic only when you need the explicit advanced surface.

Example workflows

Retrieve project context

Ask Claude Code to use search for the current architecture, then read a cited file if you need the exact implementation.

Save a durable decision

Ask Claude Code to use remember for facts that should survive the current session, such as coding standards or integration constraints.

Explore an unfamiliar repo

Ask Claude Code to use search_code for concepts, grep for exact symbols, and explore when you need to orient around the tree first.

Configuration

Example .claude/settings.local.json entry:

{
  "mcpServers": {
    "retaindb-context": {
      "command": "npx",
      "args": ["-y", "@retaindb/mcp-server@latest"],
      "env": {
        "RETAINDB_API_KEY": "rdb_...",
        "RETAINDB_PROJECT": "my-project"
      }
    }
  }
}

Best practices

1. store durable preferences with remember, not record
2. use search for stored RetainDB context and search_code for local workspace concepts
3. keep the project default scoped when Claude Code is attached to one repo or environment

Next step

• Claude Code MCP Setup
• MCP Setup
• MCP Semantic Search