@cstan0824/remembry-mcp

A Model Context Protocol server that connects AI assistants to Remembry, an AI-powered meeting knowledge base. Query meetings, manage projects, search action items, and organize your recordings — all through natural conversation with your AI assistant.

Features

• 24 MCP tools — full CRUD for projects, events, knowledge, documents, transcriptions, app launcher, and recording control
• Recording requires app — audio recording triggers the Remembry desktop app (auto-opens if not running)
• Full-text search — search across meeting titles, transcriptions, and AI-extracted knowledge
• Knowledge aggregation — pull action items, decisions, and questions across all events
• Zero configuration — auto-detects the Remembry database on your platform
• Cross-platform — works on Windows, macOS, and Linux
• Stdio transport — compatible with any MCP client (Claude, Cursor, Windsurf, etc.)

Architecture

┌─────────────────────┐       ┌──────────────────────┐
│   AI Assistant      │       │   remembry-mcp       │
│   (Claude, Cursor,  │◄─────►│   (Node.js server)   │
│    Hermes, etc.)    │  MCP  │                      │
└─────────────────────┘       └──────┬──────────┬────┘
                                     │          │
                                reads/writes   HTTP API
                                     │          │
                          ┌──────────▼───┐  ┌───▼──────────────┐
                          │  remembry    │  │  Remembry Desktop │
                          │  .sqlite3    │  │  App (Tauri)      │
                          │  (database)  │  │  • Records audio  │
                          └──────────▲───┘  │  • Transcribes    │
                                     │      │  • Manages data   │
                                     └──────┴──────────────────┘

Data access: The MCP server reads/writes the SQLite database directly — no app required for queries.

Recording: The MCP server calls the desktop app's local HTTP API (port 17890) to trigger recording. The app handles audio capture (MediaRecorder) and transcription (Gemini). The HTTP server starts on-demand and shuts down after 5 minutes of inactivity.

Installation

Prerequisites

• Node.js 18 or later
• Remembry desktop app installed and opened at least once

Quick Start

npx -y @cstan0824/remembry-mcp

This downloads and runs the latest version. No npm install required.

Configuration

Claude Code

claude mcp add remembry -- npx -y @cstan0824/remembry-mcp

Claude Desktop

Add to your Claude Desktop config file:

| Platform | Path | |----------|------| | macOS | ~/Library/Application Support/Claude/claude_desktop_config.json | | Windows | %APPDATA%\Claude\claude_desktop_config.json |

{
  "mcpServers": {
    "remembry": {
      "command": "npx",
      "args": ["-y", "@cstan0824/remembry-mcp"]
    }
  }
}

Cursor

Add to .cursor/mcp.json in your project root:

{
  "mcpServers": {
    "remembry": {
      "command": "npx",
      "args": ["-y", "@cstan0824/remembry-mcp"]
    }
  }
}

Windsurf

Add to ~/.codeium/windsurf/mcp_config.json:

{
  "mcpServers": {
    "remembry": {
      "command": "npx",
      "args": ["-y", "@cstan0824/remembry-mcp"]
    }
  }
}

VS Code (via Continue.dev or MCP extension)

Add to your VS Code MCP configuration:

{
  "servers": {
    "remembry": {
      "command": "npx",
      "args": ["-y", "@cstan0824/remembry-mcp"]
    }
  }
}

Custom Database Path

If your Remembry database is in a non-default location:

{
  "mcpServers": {
    "remembry": {
      "command": "npx",
      "args": ["-y", "@cstan0824/remembry-mcp"],
      "env": {
        "REMEMBRY_DB_PATH": "/path/to/remembry.sqlite3"
      }
    }
  }
}

Or pass it as a CLI argument:

{
  "args": ["-y", "@cstan0824/remembry-mcp", "--db-path", "/path/to/remembry.sqlite3"]
}

How It Works

Database Location

The server auto-detects the Remembry database based on your platform:

| Platform | Path | |----------|------| | Windows | %APPDATA%\com.remembry.desktop\remembry.sqlite3 | | macOS | ~/Library/Application Support/com.remembry.desktop/remembry.sqlite3 | | Linux | ~/.local/share/com.remembry.desktop/remembry.sqlite3 |

Override with the REMEMBRY_DB_PATH environment variable or --db-path argument.

Data Flow

1. Desktop App — Records audio, uploads to Gemini for transcription, extracts knowledge (summaries, action items, decisions, questions), stores everything in SQLite
2. MCP Server — Reads and writes the same SQLite database, exposes 20 tools via the MCP protocol
3. AI Assistant — Calls MCP tools to query, search, create, update, or delete meeting data

Knowledge Format

Meeting knowledge is stored per-language in the knowledge_by_language JSON column. The canonical format is EventKnowledge (schema v1):

{
  "schema_version": 1,
  "event_type": "meeting",
  "title": "Sprint Planning",
  "summary": "Discussed Q3 roadmap priorities...",
  "action_items": [
    {
      "id": "a1",
      "type": "task",
      "content": "Finalize API design",
      "assignee": "Alice",
      "due_date": "2026-06-20"
    }
  ],
  "decisions": [
    {
      "id": "d1",
      "type": "decision",
      "content": "Use REST over GraphQL for v2"
    }
  ],
  "questions": [
    {
      "id": "q1",
      "type": "question",
      "content": "Should we migrate the database?",
      "status": "open"
    }
  ],
  "key_points": [...],
  "insights": [...],
  "concepts": [...],
  "sentiment": { "overall": "positive", "important_emotions": [] }
}

Tools

Read Tools

remembry_config_status

Check database health, path, and basic statistics.

Inputs: (none)

Returns: db_path, db_exists, db_readable, projects (count), events (count), jobs (count)

remembry_list_projects

List all projects in the knowledge base.

Inputs: (none)

Returns: Array of projects with id, display_name, color, description, goals, created_at

remembry_get_project

Get details of a specific project by ID or display name (fuzzy match).

| Input | Type | Required | Description | |-------|------|----------|-------------| | project | string | yes | Project ID or display name |

Returns: Project details + event_count

remembry_list_events

List events (meetings, interviews, standups) with optional filters.

| Input | Type | Required | Description | |-------|------|----------|-------------| | project | string | no | Filter by project ID or name | | type | string | no | Filter by event type (e.g. meeting, interview) | | tag | string | no | Filter by event tag | | since | string | no | Filter events after ISO date (e.g. 2026-01-01) | | limit | number | no | Max results (default: 50) |

Returns: Array of event metadata (without full transcription)

remembry_get_event

Get full details of a specific event including transcription text.

| Input | Type | Required | Description | |-------|------|----------|-------------| | event_id | string | yes | The event/meeting ID |

Returns: Full event object with transcription, event_tags, available_languages, knowledge_languages

remembry_get_event_knowledge

Get AI-extracted knowledge for an event: summary, action items, decisions, questions, insights, concepts.

| Input | Type | Required | Description | |-------|------|----------|-------------| | event_id | string | yes | The event/meeting ID | | lang | string | no | Language code (default: en) |

Returns: Full EventKnowledge object for the specified language

remembry_search_knowledge

Aggregate action items, decisions, or questions across all events.

| Input | Type | Required | Description | |-------|------|----------|-------------| | type | enum | yes | actions, decisions, or questions | | project | string | no | Filter by project ID or name | | assignee | string | no | Filter actions by assignee name | | status | string | no | Filter questions: open, answered, partially_answered | | since | string | no | Filter events after ISO date | | lang | string | no | Language code (default: en) | | limit | number | no | Max events to scan (default: 100) |

Returns: Aggregated items with source_event, event_id, event_date

remembry_search

Full-text search across event titles, contexts, transcriptions, and knowledge content.

| Input | Type | Required | Description | |-------|------|----------|-------------| | query | string | yes | Search query | | project | string | no | Filter by project ID or name | | type | string | no | Filter by event type | | since | string | no | Filter events after ISO date | | limit | number | no | Max results (default: 10) |

Returns: Matching events ranked by relevance_score with snippet

remembry_get_jobs

List upload/processing jobs and their status.

| Input | Type | Required | Description | |-------|------|----------|-------------| | status | string | no | Filter by status: completed, failed, processing | | limit | number | no | Max results (default: 20) |

Returns: Array of upload job records

remembry_list_documents

List documents attached to a project.

| Input | Type | Required | Description | |-------|------|----------|-------------| | project_id | string | yes | Project ID | | limit | number | no | Max results (default: 50) |

Returns: Array of document metadata

remembry_open_app

Open the Remembry desktop app. Use this to start recording a meeting — the app handles audio capture and transcription via Gemini.

Inputs: (none)

Returns: app_path, message

remembry_start_recording

Start recording audio in the Remembry desktop app. Requires the Remembry app to be running — auto-opens the app if not. The terminal is free while recording.

| Input | Type | Required | Description | |-------|------|----------|-------------| | title | string | no | Meeting title (for reference) |

Returns: status, message

remembry_stop_recording

Stop the active recording in the Remembry desktop app. The app will transcribe via Gemini automatically.

Inputs: (none)

Returns: status, message

remembry_recording_status

Check if the Remembry app is running and recording status.

Inputs: (none)

Returns: status, message

Write Tools

remembry_create_project

Create a new project.

| Input | Type | Required | Description | |-------|------|----------|-------------| | name | string | yes | Project display name | | description | string | no | Project description | | color | string | no | Tailwind color class (default: bg-blue-500) | | goals | string | no | Project goals |

Returns: Created project object

remembry_update_project

Update an existing project.

| Input | Type | Required | Description | |-------|------|----------|-------------| | project_id | string | yes | Project ID to update | | name | string | yes | New display name | | description | string | no | New description | | color | string | no | New color class | | goals | string | no | New goals |

Returns: Updated project object

remembry_delete_project

Delete a project and all its associated meetings (cascade delete).

| Input | Type | Required | Description | |-------|------|----------|-------------| | project_id | string | yes | Project ID to delete |

Returns: Success message with count of deleted meetings

remembry_create_event

Create a new meeting/event record with optional transcription.

| Input | Type | Required | Description | |-------|------|----------|-------------| | project_id | string | yes | Project ID to attach to | | title | string | yes | Event title | | context | string | no | Event description | | file_type | string | no | audio, text, video (default: text) | | event_type | string | no | meeting, interview, standup, lecture (default: meeting) | | event_tags | string[] | no | Tags for the event | | transcription_text | string | no | Transcription text content | | transcription_language | string | no | Language code (default: en) | | default_language | string | no | Default language (default: en) |

Returns: Created event object

remembry_update_event

Update an existing event's metadata.

| Input | Type | Required | Description | |-------|------|----------|-------------| | event_id | string | yes | Event ID to update | | title | string | no | New title | | context | string | no | New context | | event_type | string | no | New event type | | event_tags | string[] | no | New tags (replaces existing) | | default_language | string | no | New default language |

Returns: Updated event object

remembry_delete_event

Delete a specific event by ID.

| Input | Type | Required | Description | |-------|------|----------|-------------| | event_id | string | yes | Event ID to delete |

Returns: Success message

remembry_update_event_knowledge

Write or update AI-extracted knowledge for an event in a specific language.

| Input | Type | Required | Description | |-------|------|----------|-------------| | event_id | string | yes | Event ID | | lang | string | no | Language code (default: en) | | knowledge | object | yes | Full EventKnowledge JSON object |

Returns: Success confirmation

remembry_update_meeting_transcription

Write or update the transcription text for a meeting.

| Input | Type | Required | Description | |-------|------|----------|-------------| | event_id | string | yes | Event ID | | text | string | yes | Transcription text | | language | string | no | Language code (default: en) |

Returns: Success confirmation

remembry_create_document

Create a text document attached to a project.

| Input | Type | Required | Description | |-------|------|----------|-------------| | project_id | string | yes | Project ID | | display_name | string | yes | Document name | | content | string | yes | Document text content | | mime_type | string | no | MIME type (default: text/plain) | | metadata | object | no | Optional metadata JSON |

Returns: Created document ID and name

remembry_delete_document

Delete a document by ID.

| Input | Type | Required | Description | |-------|------|----------|-------------| | document_id | string | yes | Document ID to delete |

Returns: Success message

Tool Annotations

| Tool | Read-only | Idempotent | Destructive | |------|-----------|------------|-------------| | remembry_config_status | ✅ | ✅ | ❌ | | remembry_list_projects | ✅ | ✅ | ❌ | | remembry_get_project | ✅ | ✅ | ❌ | | remembry_list_events | ✅ | ✅ | ❌ | | remembry_get_event | ✅ | ✅ | ❌ | | remembry_get_event_knowledge | ✅ | ✅ | ❌ | | remembry_search_knowledge | ✅ | ✅ | ❌ | | remembry_search | ✅ | ✅ | ❌ | | remembry_get_jobs | ✅ | ✅ | ❌ | | remembry_list_documents | ✅ | ✅ | ❌ | | remembry_open_app | ❌ | ✅ | ❌ | | remembry_start_recording | ❌ | ❌ | ❌ | | remembry_stop_recording | ❌ | ❌ | ❌ | | remembry_recording_status | ✅ | ✅ | ❌ | | remembry_create_project | ❌ | ❌ | ❌ | | remembry_update_project | ❌ | ✅ | ❌ | | remembry_delete_project | ❌ | ✅ | ✅ | | remembry_create_event | ❌ | ❌ | ❌ | | remembry_update_event | ❌ | ✅ | ❌ | | remembry_delete_event | ❌ | ✅ | ✅ | | remembry_update_event_knowledge | ❌ | ✅ | ❌ | | remembry_update_meeting_transcription | ❌ | ✅ | ❌ | | remembry_create_document | ❌ | ❌ | ❌ | | remembry_delete_document | ❌ | ✅ | ✅ |

Usage Examples

Once connected, ask your AI assistant naturally:

Searching & querying:

• "What meetings do I have?"
• "Search for meetings about the API redesign"
• "What are my pending action items?"
• "What decisions were made in the SPM project?"
• "Are there any open questions from last week's standup?"

Managing data:

• "Create a new project called Engineering Standups"
• "Add a meeting to the SPM project about requirements review"
• "Update the title of that meeting to 'Sprint 3 Planning'"

Analysis:

• "Summarize the last meeting"
• "What action items are assigned to John?"
• "What topics keep coming up across meetings?"

Recording (via desktop app):

• "Record a meeting called Sprint Planning" → opens app, starts recording
• "Stop recording" → stops and transcribes
• "What new meetings do I have?" → lists recent meetings

Development

# Clone the repository
git clone https://github.com/kongjiyu/remembry-app.git
cd remembry-app/mcp

# Install dependencies
npm install

# Build
npm run build

# Run in development mode
npm run dev

Releasing

# Bump version
npm version patch   # 0.1.0 → 0.1.1
npm version minor   # 0.1.0 → 0.2.0
npm version major   # 0.1.0 → 1.0.0

# Publish
npm publish --access public

Or push a git tag to trigger CI auto-publish:

git tag v0.2.0
git push --tags

Troubleshooting

"Database not found" error

• Make sure the Remembry desktop app is installed and has been opened at least once
• Verify the database path matches your platform (see Database Location)
• Use the remembry_config_status tool to check DB accessibility

Tools not appearing after config

• Restart your MCP client after adding the configuration
• Verify Node.js is installed: node --version
• Test the server directly: npx -y @cstan0824/remembry-mcp

Permission errors on database

• The MCP server opens the database in WAL mode (concurrent read/write safe)
• Ensure no other process has an exclusive lock on the file
• Try closing the Remembry desktop app before querying

Links

• Remembry Desktop App
• Model Context Protocol
• MCP Specification
• npm Package

License

MIT