dialogue-db-mcp

MCP (Model Context Protocol) server for DialogueDB - persist, search, and resume AI conversations across devices.

Features

• Save Sessions - Persist your AI agent conversations to the cloud
• Resume Anywhere - Continue conversations from any device
• Search History - Find relevant past conversations by content
• Threading - Create branching conversations for exploring alternatives
• State Management - Persist conversation context and preferences

Installation

With Goose

goose configure
# → Add Extension
# → npm package
# → dialogue-db-mcp

Or add to your Goose config directly:

# ~/.config/goose/config.yaml
extensions:
  dialoguedb:
    type: stdio
    cmd: npx
    args: ["dialogue-db-mcp"]
    env:
      DIALOGUEDB_API_KEY: "your-api-key"

With Claude Desktop

Add to your Claude Desktop config:

{
  "mcpServers": {
    "dialoguedb": {
      "command": "npx",
      "args": ["dialogue-db-mcp"],
      "env": {
        "DIALOGUEDB_API_KEY": "your-api-key"
      }
    }
  }
}

Standalone

npm install -g dialogue-db-mcp

# Run directly
DIALOGUEDB_API_KEY=your-key dialoguedb-mcp

Available Tools

dialoguedb_save

Save the current conversation to DialogueDB.

You: Save this session to DialogueDB with label "Auth refactoring"

Goose: ✓ Session saved. Resume ID: conv_abc123

Parameters:

• messages (required) - Array of conversation messages
• label - Human-readable label
• id - Custom session ID (auto-generated if not provided)
• state - Conversation state/context to persist
• tags - Tags for categorization
• metadata - Additional metadata

dialoguedb_resume

Load a previously saved session.

You: Resume session conv_abc123

Goose: Loaded session with 15 messages. Last modified 2 hours ago.
       Continuing from: "Let's implement the OAuth flow..."

Parameters:

• id (required) - Session ID to resume
• loadMessages - Whether to load message history (default: true)

dialoguedb_search

Search across all saved sessions.

You: Search my sessions for "authentication middleware"

Goose: Found 3 relevant sessions:
       1. "Auth refactoring" (conv_abc123) - 2 days ago
       2. "Security review" (conv_def456) - 1 week ago
       3. "API design" (conv_ghi789) - 2 weeks ago

Parameters:

• query (required) - Search query
• limit - Maximum results (default: 10)
• searchMessages - Also search message content (default: false)

dialoguedb_list

List saved sessions with optional filters.

Parameters:

• limit - Maximum results (default: 20)
• order - Sort order: "asc" or "desc" (default: desc)
• startDate - Filter by creation date (ISO format)
• endDate - Filter by creation date (ISO format)

dialoguedb_add_message

Add a message to an existing session.

Parameters:

• sessionId (required) - Session to add to
• role (required) - Message role (user, assistant, system, tool)
• content (required) - Message content
• name - Optional sender name

dialoguedb_update_state

Update conversation state/context.

Parameters:

• sessionId (required) - Session to update
• state (required) - State object
• merge - Merge with existing state (default: true)

dialoguedb_create_thread

Create a branching conversation from an existing session.

Parameters:

• parentId (required) - Parent session ID
• label - Label for the new thread
• tags - Tags for the new thread

Environment Variables

| Variable | Description | Required | |----------|-------------|----------| | DIALOGUEDB_API_KEY | Your DialogueDB API key | Yes | | DIALOGUEDB_BASE_URL | Custom API endpoint (for self-hosted) | No |

Development

# Install dependencies
npm install

# Build
npm run build

# Run with MCP Inspector for debugging
npm run inspect

# Watch mode for development
npm run dev

Use Cases

Cross-Device Continuity

Start a conversation on your laptop, continue on your desktop:

# On laptop
You: Save this session for later

# On desktop
You: Resume my session about the auth refactoring

Team Collaboration

Share conversation context with teammates:

You: Save this session with tag "team-review"

# Teammate can then:
You: Search sessions tagged "team-review"

CI/CD Integration

Persist automated agent runs for analysis:

# GitHub Actions
- name: Run Analysis
  env:
    DIALOGUEDB_API_KEY: ${{ secrets.DIALOGUEDB_API_KEY }}
  run: |
    goose run --with-extension dialogue-db-mcp \
      "Analyze this PR and save findings to DialogueDB"

Context Preservation

Never lose important context when hitting token limits:

You: Save the current state before we compact

# After compaction
You: What was our previous discussion about error handling?
# Search your saved sessions to recover context

License

MIT

Links

• DialogueDB
• Documentation
• MCP Specification