Getting Started
Prerequisites
No other external services or dependencies are required. Zakira uses SQLite for storage and an embedded ONNX model for semantic search.
1. Download the ONNX Model
The semantic search feature requires the all-MiniLM-L6-v2 ONNX model. Download it using one of the provided scripts:
PowerShell:
./scripts/download-model.ps1
Bash:
./scripts/download-model.sh
This downloads:
all-MiniLM-L6-v2.onnx(~90 MB) – the sentence transformer modelvocab.txt(~230 KB) – the WordPiece vocabulary file
Both files are placed in src/Zakira.Exchange.Core/Models/.
Note: The model files are included in the build output automatically. They are gitignored since they are large binary files downloaded from HuggingFace.
2. Build
dotnet build Zakira.Exchange.slnx
3. Use as a CLI Tool
Create, list, and search memories directly from the command line:
# Create a memory entry
dotnet run --project src/Zakira.Exchange.Cli -- create general my-key --data "Hello, world!"
# List all entries
dotnet run --project src/Zakira.Exchange.Cli -- list
# Search using natural language
dotnet run --project src/Zakira.Exchange.Cli -- search "greeting"
# List all categories
dotnet run --project src/Zakira.Exchange.Cli -- categories
See the CLI Reference for the complete command documentation.
4. Start as an MCP Server
Launch Zakira as an MCP server for AI agent integration:
# Start with default settings
dotnet run --project src/Zakira.Exchange.Cli -- mcp
# Start in read-only mode
dotnet run --project src/Zakira.Exchange.Cli -- mcp --access-mode read-only
# Start locked to a single category with a custom database
dotnet run --project src/Zakira.Exchange.Cli -- mcp --category notes --db ./notes.db
See MCP Server for detailed setup instructions for Claude Desktop, VS Code, and Cursor.
Install as a Global Tool (Optional)
You can also pack and install Zakira as a .NET global tool:
dotnet pack src/Zakira.Exchange.Cli/Zakira.Exchange.Cli.csproj
dotnet tool install --global --add-source ./src/Zakira.Exchange.Cli/bin/Release Zakira.Exchange
Then use zakira directly:
zakira create general my-key --data "Hello, world!"
zakira search "greeting"
zakira mcp
Next Steps
- CLI Reference – Learn all available commands and options
- MCP Server – Configure Zakira for your AI workflow
- Configuration – Customize access modes, database paths, and more