Installation¶

Current release: v2.3.0 (changelog)

This guide shows how to configure and run Sentinel DV locally, then connect it to an MCP client (e.g. Claude Desktop).

1. Install¶

From MCP Registry (recommended)¶

Use server name io.github.kiranreddi/sentinel-dv in an MCP registry–aware client, or run with uvx:

uvx --from sentinel-dv sentinel-dv-server --config /absolute/path/to/config.yaml

From PyPI¶

pip install "sentinel-dv>=2.3.0"

Note: Do not install PyPI 1.0.0 (broken wheel). Use sentinel-dv>=2.3.0 for VCS/Questa/Cadence fixtures, assertion/coverage intelligence, VCD/JSON waveforms, and all 28 MCP tools. Verified flow: install → sentinel-dv-index --index-all → sentinel-dv-server → MCP tools.

For development:

pip install -e ".[dev]"

2. Create config.yaml (required)¶

Sentinel DV requires a config file before indexing or serving MCP tools. There is no automatic demo/ fallback.

Copy the provided template and update:

cp config.example.yaml config.yaml

You may also use config.yml in the working directory, or set SENTINEL_DV_CONFIG / pass --config with any path you choose.

Edit config.yaml to set:

• artifact_roots: directories that contain your verification artifacts (UVM *.log, cocotb JUnit XML like results.xml / junit.xml)
• index.path: where the DuckDB database should be written
• adapters: enable/disable parsers (booleans under uvm, cocotb, assertions, coverage, waveform_summary)
• For waveforms: set waveform_summary: true and add *.wave.json and/or *.vcd files whose test_name matches indexed tests (see Waveform summaries and demo/verilator_counter/)

3. Index artifacts (required)¶

python -m sentinel_dv.indexing.indexer --config config.yaml --index-all

This scans the configured artifact_roots and rebuilds the index.

4. Start the MCP server (stdio)¶

python -m sentinel_dv.server --config config.yaml

Alternative: set the config path once:

export SENTINEL_DV_CONFIG=/absolute/path/to/config.yaml
python -m sentinel_dv.server

5. Connect via Claude Desktop¶

Add Sentinel DV to your Claude Desktop MCP config:

{
  "mcpServers": {
    "sentinel-dv": {
      "command": "python",
      "args": [
        "-m",
        "sentinel_dv.server",
        "--config",
        "/absolute/path/to/config.yaml"
      ]
    }
  }
}

6. Connect via Cline (VS Code)¶

Update .vscode/mcp.json:

{
  "mcpServers": {
    "sentinel-dv": {
      "command": "python",
      "args": [
        "-m",
        "sentinel_dv.server",
        "--config",
        "${workspaceFolder}/config.yaml"
      ]
    }
  }
}

Notes¶

• Sentinel DV exposes MCP tools over stdio (there are currently no documented HTTP endpoints).
• If the server reports INDEX_NOT_READY, run the indexing step again.