Use Kshana inside your AI agent

Kshana ships an MCP server, kshana-mcp, that exposes the validated engine as agent tools. MCP is the shared plugin protocol, so the same server plugs into Claude Code, Claude Desktop, Codex, Cursor, VS Code, Windsurf, and JetBrains — the agent calls the real, externally-validated engine instead of guessing the math.

Tools exposed: run_scenario, list_scenario_kinds, validate_scenario, export_sp3, export_omm — each a thin, faithful wrapper over a public kshana::api function.

1. Install the server (once)

# crates.io — a Rust toolchain builds the prebuilt source onto your PATH (~/.cargo/bin):
cargo install kshana-mcp

# …or Docker / OCI — no Rust toolchain, amd64 + Apple Silicon:
docker run --rm -i ghcr.io/ashfordeou/kshana-mcp

Every host below runs the same stdio server. Where a host wants an absolute path, it's usually ~/.cargo/bin/kshana-mcp (run command -v kshana-mcp to confirm).

2. Wire it into your host

Claude Code

Fastest — one command:

claude mcp add kshana -- kshana-mcp

Or install the plugin (bundles the server, adds a /kshana-run command):

/plugin marketplace add ashfordeOU/kshana
/plugin install kshana@ashforde

Docker instead of the binary:

claude mcp add kshana -- docker run --rm -i ghcr.io/ashfordeou/kshana-mcp

Claude Desktop

Edit claude_desktop_config.json (Settings → Developer → Edit Config):

{
  "mcpServers": {
    "kshana": { "command": "kshana-mcp", "args": [] }
  }
}

Codex CLI

Add to ~/.codex/config.toml:

[mcp_servers.kshana]
command = "kshana-mcp"
args = []

Cursor

.cursor/mcp.json (project) or ~/.cursor/mcp.json (global):

{
  "mcpServers": {
    "kshana": { "command": "kshana-mcp", "args": [] }
  }
}

VS Code (Copilot agent mode / Continue)

.vscode/mcp.json:

{
  "servers": {
    "kshana": { "type": "stdio", "command": "kshana-mcp", "args": [] }
  }
}

Windsurf

~/.codeium/windsurf/mcp_config.json:

{
  "mcpServers": {
    "kshana": { "command": "kshana-mcp", "args": [] }
  }
}

JetBrains AI Assistant / Junie

Settings → Tools → AI Assistant → MCP → add a stdio server with command kshana-mcp.

3. Try it

Ask the agent something the engine is validated for, e.g.:

"Use kshana: run a clock-holdover scenario with an optical clock through a 1-hour GNSS outage and report the p95 timing error and availability."

The agent calls list_scenario_kinds → builds the TOML → run_scenario, and reports figures of merit with a scenario + seed + engine version provenance line — reproducible, not guessed.

Notes

• Absolute paths: if a host can't find kshana-mcp, give the full path from command -v kshana-mcp (typically ~/.cargo/bin/kshana-mcp).
• Docker form: replace "command": "kshana-mcp", "args": [] with "command": "docker", "args": ["run", "--rm", "-i", "ghcr.io/ashfordeou/kshana-mcp"].
• Registry: the server is also published to the MCP registry (io.github.ashfordeOU/kshana-mcp) and listed on Glama.