Updated 2026-10-02

OpenClaw

OpenClaw manages client-side MCP servers through its MCP registry and Control UI. Do not use MCPorter configuration for OpenClaw-managed servers.

Get the exact values with the CLI

LyraShield prints the connection values, then OpenClaw stores them through its own MCP UI or CLI. Run npx lyrashield install openclaw to retrieve the local command values.

npx lyrashield install openclaw

Add the server in OpenClaw

In the Control UI, open Settings → MCP, add a server and enter the values below. Alternatively use openclaw mcp add for a local server or openclaw mcp set for a remote server.

Use this command in a terminal or enter the same values in the agent's MCP settings UI. Recommended: run npx -y lyrashield@0.2.14 login --oauth once; the local MCP server reads the selected workspace credential from ~/.lyrashield/credentials.json. If you use an API key instead, set it in your private process environment before starting the client.

LYRASHIELD_API_URL=https://app.lyrashieldai.com npx -y @lyrashield/mcp@0.2.12

npx -y lyrashield@0.2.14 install openclaw

Important gotchas

• OpenClaw manages client-side servers with `openclaw mcp add`, `set` and `configure` or in its Control UI at /settings/mcp. Do not use mcporter configuration for OpenClaw-managed servers.
• Local entries use `command` and repeated `--arg` flags. For Streamable HTTP, use `transport: "streamable-http"`; then run `openclaw mcp doctor --probe` for a live tool-list check.

Remote OAuth alternative

OpenClaw's current MCP client also supports Streamable HTTP and OAuth. Save the remote endpoint as a separate client-managed connection, complete openclaw mcp login and use openclaw mcp probe to confirm a live tool list before switching from stdio.

Remote Streamable-HTTP setup. Complete OAuth in the client after adding the server. No API key belongs in this config.

{
  "mcpServers": {
    "lyrashield": {
      "transport": "streamable-http",
      "url": "https://app.lyrashieldai.com/api/mcp"
    }
  }
}

Skill definition (optional)

Create a skill.md file in your OpenClaw skills directory to teach the agent when to invoke LyraShield automatically. For example:

# LyraShield Security Checks

Run LyraShield checks before submitting a pull request or after applying a fix.

## When to use
- Before creating a PR: run check_diff on the staged changes
- After applying a security fix: run verify_fix to confirm resolution
- When reviewing dependency updates: run supply chain checks

## How to invoke
Use the lyrashield MCP tools available in the tool registry.

Verify the connection

Run openclaw mcp doctor --probe after saving. It checks the saved configuration, connects to the server and lists the available tools.