Caution
PROJECT DISCONTINUED — The replacement solution is being developed at mnemo-mcp.
Black box for all your AI conversations.
EchoVault extracts, indexes, and searches chat history from 12+ AI coding tools — ensuring you never lose valuable insights. Works as a desktop app, CLI, or MCP server for AI assistants.
- 12 Source Extractors: VS Code Copilot, Cursor, Cline, Continue.dev, JetBrains AI, Zed, Antigravity, Gemini CLI, Claude Code, Aider, Codex, OpenCode
- Hybrid Search: Vector semantic search + FTS5 keyword search with RRF fusion
- MCP Server: Expose your vault to Claude Desktop, Copilot, Cursor, and other AI assistants
- Cloud Sync: Auto-sync with Google Drive via Rclone
- Desktop App: Mini window with system tray, background sync, auto-update
- Embedding Presets: Built-in Ollama/OpenAI support — no external proxy needed
- Cross-platform: Windows, Linux, macOS
- Future-proof: Stores raw files without transformation
IDE Sources (12) EchoVault Pipeline AI Assistants
+-----------------+ +-------------------------+ +----------------+
| VS Code Copilot | | | | Claude Desktop |
| Cursor | --> | Extract --> Parse --> | | VS Code Copilot|
| Cline | | Embed --> Index --> | --> | Cursor |
| Continue.dev | | Search (Hybrid) | | Any MCP Client |
| JetBrains AI | | | +----------------+
| Zed | +----+----+----+----------+
| Gemini CLI | | | |
| Claude Code | vault.db index.db embeddings.db
| Aider | |
| Codex | Google Drive (Rclone sync)
| OpenCode |
| Antigravity |
+-----------------+
Warning
EchoVault accesses your AI chat history. This data may contain:
- Code snippets and file paths
- API keys or secrets mentioned in conversations
- Personal information
Cloud sync is optional. All data stays local unless you explicitly connect Google Drive.
The MCP server exposes your vault to AI assistants via 2 tools:
| Tool | Description |
|---|---|
vault |
Unified interface: list, search (FTS5), read, semantic_search (hybrid) |
help |
On-demand documentation (saves tokens — only called when needed) |
Quick install (recommended):
# Linux/macOS
curl -fsSL https://raw.githubusercontent.com/n24q02m/EchoVault/main/install-cli.sh | bash# Windows (PowerShell)
irm https://raw.githubusercontent.com/n24q02m/EchoVault/main/install-cli.ps1 | iexTip
The full install script (Desktop App) also includes the CLI.
Only use install-cli if you need CLI/MCP only (e.g., servers, CI, headless).
Build from source:
git clone https://github.com/n24q02m/EchoVault.git
cd EchoVault
cargo build -p echovault-cli --release
# Binary at: target/release/echovault-cli# Extract sessions from all detected IDEs
echovault-cli extract
# Parse raw files into clean Markdown
echovault-cli parse
# (Optional) Build embedding index for semantic search
# Requires Ollama running locally, or set OpenAI API key in config
echovault-cli embedClaude Desktop (claude_desktop_config.json):
{
"mcpServers": {
"echovault": {
"command": "echovault-cli",
"args": ["mcp"]
}
}
}VS Code Copilot (.vscode/mcp.json):
{
"servers": {
"echovault": {
"type": "stdio",
"command": "echovault-cli",
"args": ["mcp"]
}
}
}Cursor (~/.cursor/mcp.json):
{
"mcpServers": {
"echovault": {
"command": "echovault-cli",
"args": ["mcp"]
}
}
}Note
The install scripts add echovault-cli to your PATH automatically.
If you built from source, use the full path: "command": "/path/to/target/release/echovault-cli"
Once configured, your AI assistant can:
"List my recent AI chat sessions"
→ vault(action="proxy.php?url=https%3A%2F%2Fwww.github.com%2Flist", limit=20)
"Find sessions about authentication"
→ vault(action="proxy.php?url=https%3A%2F%2Fwww.github.com%2Fsearch", query="authentication")
"Show me the full conversation from that Cursor session"
→ vault(action="proxy.php?url=https%3A%2F%2Fwww.github.com%2Fread", source="cursor", session_id="abc123")
"Find code related to database migrations"
→ vault(action="proxy.php?url=https%3A%2F%2Fwww.github.com%2Fsemantic_search", query="database migration setup")
Installs both the Desktop App and CLI (including MCP server).
Linux/macOS:
curl -fsSL https://raw.githubusercontent.com/n24q02m/EchoVault/main/install.sh | bashWindows (PowerShell):
irm https://raw.githubusercontent.com/n24q02m/EchoVault/main/install.ps1 | iexTip
Need CLI only (no desktop app)? See MCP Server > Install CLI.
Download from Releases:
| Platform | File |
|---|---|
| Windows (x64) | EchoVault_x.x.x_x64-setup.exe |
| macOS (Intel) | EchoVault_x.x.x_x64.dmg |
| macOS (Apple Silicon) | EchoVault_x.x.x_aarch64.dmg |
| Linux (Debian/Ubuntu) | EchoVault_x.x.x_amd64.deb |
| Linux (Fedora/RHEL) | EchoVault-x.x.x-1.x86_64.rpm |
| Linux (Universal) | EchoVault_x.x.x_amd64.AppImage |
The desktop app provides a GUI for extraction, parsing, embedding, search, cloud sync, and settings.
Pre-built CLI binaries are also attached to each release:
| Platform | File |
|---|---|
| Linux (x64) | echovault-cli-linux-x64 |
| macOS (Intel) | echovault-cli-macos-x64 |
| macOS (Apple Silicon) | echovault-cli-macos-arm64 |
| Windows (x64) | echovault-cli-windows-x64.exe |
For servers, CI environments, or headless usage. Install via quick install or download from Releases.
echovault-cli <COMMAND>
Commands:
auth Authenticate with Google Drive
sync Sync vault (pull -> extract -> push)
extract Extract sessions from all detected IDEs
parse Parse raw sessions into clean Markdown
embed Build embedding index for semantic search
search Semantic search across embedded conversations
mcp Start MCP server on stdio
intercept Start interceptor proxy for API traffic capture
status Show current status (auth, sync, vault info)# Full pipeline: extract -> parse -> embed -> ready for search/MCP
echovault-cli extract
echovault-cli parse
echovault-cli embed
# Quick search
echovault-cli search "how to setup fastapi middleware" --limit 5
# Cloud sync (requires auth first)
echovault-cli auth
echovault-cli sync| Source | IDE Support | Storage Format |
|---|---|---|
vscode-copilot |
VS Code, VS Code Insiders | JSON/JSONL per workspace |
cline |
VS Code, Cursor | JSON tasks in globalStorage |
continue-dev |
VS Code, JetBrains | JSON sessions in ~/.continue/ |
| Source | Description | Storage Format |
|---|---|---|
cursor |
Cursor AI Editor | SQLite database |
jetbrains |
IntelliJ, PyCharm, WebStorm, GoLand, etc. | XML workspace files |
zed |
Zed Editor | SQLite (zstd compressed) |
antigravity |
Google Antigravity IDE | Protobuf + Markdown |
| Source | Description | Storage Format |
|---|---|---|
gemini-cli |
Google Gemini CLI | JSON sessions |
claude-code |
Claude Code (Anthropic) | JSONL sessions |
aider |
Aider AI assistant | Markdown history |
codex |
OpenAI Codex CLI | JSONL rollout |
opencode |
OpenCode terminal AI | JSON sessions |
EchoVault supports hybrid search combining vector similarity and keyword matching.
Configure via Desktop App (Settings) or ~/.config/echovault/echovault.toml:
| Preset | API Base | Default Model | API Key Required |
|---|---|---|---|
| Ollama (default) | http://localhost:11434/v1 |
nomic-embed-text |
No |
| OpenAI | https://api.openai.com/v1 |
text-embedding-3-small |
Yes |
| Custom | Any OpenAI-compatible endpoint | User-defined | Depends |
# ~/.config/echovault/echovault.toml
[embedding]
preset = "ollama"
api_base = "http://localhost:11434/v1"
model = "nomic-embed-text"
# api_key = "" # Only needed for OpenAI/custom- FTS5 keyword search — SQLite full-text search on chunks (BM25 ranking)
- Vector similarity — Cosine similarity on embeddings
- RRF fusion — Reciprocal Rank Fusion merges both result sets (alpha=0.6 vector bias)
- mise (auto-installs Rust, Node.js, pnpm, uv)
git clone https://github.com/n24q02m/EchoVault.git
cd EchoVault
mise run setup# Rust
cargo build --workspace # Debug build (all crates)
cargo test --workspace # Run tests (36 tests)
cargo clippy --workspace # Lint
cargo fmt --all # Format
# Frontend (apps/web)
cd apps/web
pnpm dev # Dev server with HMR
pnpm build # Production build
pnpm lint # Biome lint
# Desktop app
cargo tauri dev # Development mode
cargo tauri build # Production buildEchoVault/
apps/
core/ # Core library (extractors, parsers, embedding, MCP, storage)
cli/ # CLI binary (echovault-cli)
tauri/ # Desktop app (Tauri + React)
web/ # Frontend (React + TypeScript + Tailwind)
scripts/ # Build scripts (download rclone, setup dev, etc.)
| Feature | Description | Used By |
|---|---|---|
embedding |
Vector embeddings + hybrid search | CLI, Tauri |
mcp |
MCP server (rmcp + stdio transport) | CLI |
interceptor |
MITM proxy for API traffic capture | CLI, Tauri |
See CONTRIBUTING.md for development guidelines.
MIT - See LICENSE for details.