AI Session Bridge

Cross-tool AI conversation history reader. Enables any AI coding agent to access conversation history from other AI tools in the same workspace.

Python License: MIT Tests codecov

Overview

AI Session Bridge is an MCP (Model Context Protocol) server and CLI that allows AI coding assistants to read conversation history from multiple tools. Instead of starting fresh every time you switch between VS Code Copilot, Cursor, or Rovodev, your AI agents can now access the full context of what you've been working on.

┌─────────────────────────────────────────────────────────────────────┐
│                         Your Workspace                               │
│  /Users/you/dev/my-project                                          │
└─────────────────────────────────────────────────────────────────────┘
                                    │
            ┌───────────────┬───────┼───────┬───────────────┐
            ▼               ▼       ▼       ▼
    ┌───────────┐   ┌───────────┐  ┌───────────┐
    │ Copilot   │   │ Cursor    │  │ Rovodev   │
    │ Sessions  │   │ Sessions  │  │ Sessions  │
    └─────┬─────┘   └─────┬─────┘  └─────┬─────┘
          └───────────────┴───────┬───────┴───────
                                  ▼
                  ┌───────────────────────────┐
                  │   ai-session-bridge       │
                  │   MCP Server + CLI        │
                  │                           │
                  │  Tools:                   │
                  │  - list_sessions          │
                  │  - get_session            │
                  │  - search_sessions        │
                  │  - get_recent_context     │
                  └───────────────────────────┘

Features

• 📚 Unified Session Reading: Read conversations from VS Code Copilot, Cursor, and Rovodev
• 🔍 Full-Text Search: Fast search across all conversations using SQLite FTS5
• 🚀 MCP Server: Expose session history via Model Context Protocol
• 💻 CLI Tool: List, search, and export sessions from the command line
• 🔒 Security: Automatic filtering of API keys, secrets, and sensitive data
• ⚡ Performance: Smart caching and indexing for instant access
• 🎯 Workspace-Aware: Automatically matches sessions to your current workspace

Screenshots

List Sessions Across All Tools

View all your AI conversations in a clean, organized table format:

Workspace-Aware Search

Search respects your current workspace - find exactly what you need:

Session Details

View full conversation history with proper formatting:

Installation

Production Use (Recommended)

Using pipx (isolated, globally available):

git clone https://github.com/frknyldz/ai-session-bridge
cd ai-session-bridge
pipx install .

This installs ai-session-bridge globally while keeping it isolated from your system Python.

From PyPI (when published):

pipx install ai-session-bridge

Development Setup

Quick Setup (automated):

git clone https://github.com/frknyldz/ai-session-bridge
cd ai-session-bridge
./setup.sh

The setup script will:

• ✅ Create a virtual environment
• ✅ Install all dependencies (editable mode)
• ✅ Set up pre-commit hooks
• ✅ Create config directories
• ✅ Test the installation
• ✅ Show both venv and pipx installation options

Manual Setup:

git clone https://github.com/frknyldz/ai-session-bridge
cd ai-session-bridge
python3 -m venv venv
source venv/bin/activate
pip install -e ".[dev]"

Quick Start

CLI Usage

# List recent sessions
ai-session-bridge list

# List sessions with filters
ai-session-bridge list --tool copilot --limit 5

# Show full conversation
ai-session-bridge show <session-id> --tool copilot

# Search across all sessions
ai-session-bridge search "bug fix"

# Clear cache
ai-session-bridge cache clear

MCP Server

Add to your AI tool's MCP configuration:

If you installed with pipx (recommended):

VS Code Copilot (settings.json):

{
  "servers": {
    "ai-session-bridge": {
      "command": "/Users/YOUR_USERNAME/.local/bin/ai-session-bridge",
      "args": ["serve"]
    }
  }
}

Cursor/Rovodev (~/.config/mcp/mcp.json):

{
  "mcpServers": {
    "ai-session-bridge": {
      "command": "/Users/YOUR_USERNAME/.local/bin/ai-session-bridge",
      "args": ["serve"]
    }
  }
}

If you're using venv (development):

Use the full path to your venv binary:

{
  "mcpServers": {
    "ai-session-bridge": {
      "command": "/absolute/path/to/ai-session-bridge/venv/bin/ai-session-bridge",
      "args": ["serve"]
    }
  }
}

Tip: Run ./setup.sh and it will show you the exact path to use!

After PyPI Release:

Use uvx to automatically download and run:

{
  "mcpServers": {
    "ai-session-bridge": {
      "command": "uvx",
      "args": ["ai-session-bridge", "serve"]
    }
  }
}

MCP Tools

Once configured, your AI assistant will have access to these tools:

list_sessions

List all AI sessions for the current workspace.

{
  "workspace_path": "/path/to/workspace",
  "tool_filter": ["copilot", "rovodev"],
  "limit": 10
}

get_session

Get the complete conversation from a specific session.

{
  "session_id": "abc123",
  "tool": "copilot",
  "workspace_path": "/path/to/workspace"
}

search_sessions

Search for keywords across all sessions.

{
  "query": "authentication bug",
  "workspace_path": "/path/to/workspace",
  "limit": 20
}

get_recent_context

Get summarized context from recent sessions (useful for "continue where I left off").

{
  "workspace_path": "/path/to/workspace",
  "max_sessions": 3,
  "max_messages": 20
}

Configuration

Create ~/.config/ai-session-bridge/config.yaml:

# Scope: which sessions to include
scope:
  mode: current_workspace  # current_workspace, all_workspaces, specified
  workspaces: []  # Only used if mode is "specified"

# Which tools to read from
tools:
  copilot: true
  cursor: true
  rovodev: true

# Cache settings
cache:
  enabled: true
  max_age_days: 30

# Search index
search:
  enabled: true
  use_fts: true  # Use SQLite FTS5 for fast search

# Security: filter sensitive content
filtering:
  enabled: true
  patterns:
    - "(?i)(api[_-]?key|apikey)[\"']?\\s*[:=]\\s*[\"']?[a-zA-Z0-9_-]{20,}"
    - "(?i)(secret|password|token)[\"']?\\s*[:=]\\s*[\"']?[^\\s\"']{8,}"
  replacement: "[REDACTED]"

# MCP server settings
mcp:
  max_context_messages: 50
  max_context_sessions: 5

Supported Tools

Tool	Location	Format	Status
VS Code Copilot	~/Library/Application Support/Code/User/workspaceStorage/{hash}/chatSessions/*.json	copilot_v3	✅ Supported
Cursor	~/Library/Application Support/Cursor/User/globalStorage/state.vscdb (new) ~/Library/Application Support/Cursor/User/workspaceStorage/{hash}/state.vscdb (old)	SQLite	✅ Supported
Rovodev	~/.rovodev/sessions/{uuid}/	rovodev_v1	✅ Supported

How It Works

Session Discovery

1. VS Code Copilot: Scans workspace directories and reads workspace.json files to match workspace paths
2. Cursor: Reads from global SQLite database (state.vscdb) containing Composer conversations
3. Rovodev: Reads workspace_path from metadata.json in each session file

Caching & Performance

• Session Cache: Stores metadata to avoid re-parsing unchanged files
• Search Index: SQLite FTS5 for instant full-text search
• Lazy Loading: Only loads full session content when needed

Security

Automatically filters:

• API keys and tokens
• Passwords and secrets
• Private keys
• AWS credentials
• GitHub/Slack tokens

Examples

Example 1: Continue Previous Work

User: "Continue where we left off"

AI uses get_recent_context to fetch the last few sessions and their messages, understanding what was being worked on.

Example 2: Find Related Discussions

User: "Did we discuss authentication in any previous conversations?"

AI uses search_sessions with query "authentication" to find relevant messages across all tools.

Example 3: Review Past Solution

User: "Show me the session where we fixed the database connection issue"

AI uses search_sessions to find the session, then get_session to retrieve the full conversation.

Development

# Clone repository
git clone https://github.com/frknyldz/ai-session-bridge
cd ai-session-bridge

# Install development dependencies
pip install -e ".[dev]"

# Install pre-commit hooks
pre-commit install

# Run tests
pytest

# Run linting
ruff check .

# Auto-fix linting issues
ruff check --fix .

# Format code
ruff format .

See CONTRIBUTING.md for detailed development guidelines.

Roadmap

• Session tagging and categorization
• Auto-summary using LLM
• Web UI for browsing sessions
• Session sync across machines
• Support for more AI tools (Continue, Windsurf, etc.)
• Export to various formats (PDF, HTML)

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

MIT License - see LICENSE for details.

Author

Furkan Yıldız

Acknowledgments

• Inspired by the need for better context continuity across AI tools
• Built on the Model Context Protocol