Connecting Slashr to Your AI Workflow via MCP

Slashr now has a Model Context Protocol (MCP) server. This means any MCP-compatible AI tool -- Claude Code, Claude Desktop, or custom agents -- can query live validator incident data, scan results, and network summaries directly.

No scraping. No CSV exports. Just structured, real-time data flowing into your AI workflow.

This post covers what the MCP server exposes, how to get an API key, and how to wire it into Claude Code in about 60 seconds.

What's Available

The Slashr MCP server at mcp.slashr.dev exposes six tools:

Every tool accepts chain-specific parameters and returns structured JSON. Responses include validator identities (resolved to operator names where possible), severity scores, timestamps, and deep links back to slashr.dev.

Getting an API Key

Keys are self-service. No sign-up, no approval process. You can generate one in two ways:

Option 1: From the website

Head to slashr.dev/developers and click "Generate Key". Your key appears immediately.

Option 2: Via the API

curl -X POST https://mcp.slashr.dev/mcp/keys/generate \
  -H "Content-Type: application/json" -d '{}'

You'll get back:

{
  "key": "slashr_...",
  "docs": "https://slashr.dev/developers",
  "mcp_url": "https://mcp.slashr.dev/mcp"
}

One key per IP per day, no auth required. Keys are Bearer tokens -- you'll pass them in the Authorization header on every request. Rate limiting is enforced server-side, and the connection pool is capped to keep things stable for everyone.

Adding Slashr to Claude Code

Claude Code supports MCP servers natively. One command, and you're connected.

Open your terminal and run:

claude mcp add slashr \
  --transport http \
  https://mcp.slashr.dev/mcp \
  --header "Authorization: Bearer slashr_xxxxxxxxxxx"

Replace slashr_xxxxxxxxxxx with the key you generated above. That's it -- Claude Code will now have access to all six Slashr tools.

To verify it's connected, ask Claude Code something like:

What are the worst offending validators on Solana this week?

or:

Check the delegation health for this wallet: <your-solana-address>

Claude will call the appropriate Slashr MCP tool, get live data, and respond with the results.

Adding to Claude Desktop

For Claude Desktop (the app), you'll need to edit your MCP configuration file. The location depends on your OS:

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json
• Linux: ~/.config/Claude/claude_desktop_config.json

Add the Slashr server to your mcpServers configuration:

{
  "mcpServers": {
    "slashr": {
      "transport": "http",
      "url": "https://mcp.slashr.dev/mcp",
      "headers": {
        "Authorization": "Bearer slashr_xxxxxxxxx"
      }
    }
  }
}

Restart Claude Desktop after saving the file. The Slashr tools will appear in the tool list.

Example Prompts

Once connected, here are some things you can ask:

For delegators:

• "Check the delegation health for wallet HN7cAB..."
• "Are any of my validators repeat offenders?"
• "Which Solana validators have the worst track record this month?"

For operators:

• "Show me the incident history for validator 5ZWgXc..."
• "What did the last infrastructure scan find on my node?"
• "Give me a network summary for Cosmos -- how many jailing events this week?"

For researchers and analysts:

• "Who are the worst offenders across all chains in the last 30 days?"
• "Compare incident rates between Solana and Sui this month"
• "Pull scan results for the top 5 worst Solana validators -- any common CVEs?"

The MCP server returns structured data, so Claude can aggregate, compare, and reason over the results. Ask follow-up questions, request charts, or pipe the data into reports -- it's all live.

Security & Rate Limits

A few notes on how the MCP server is secured:

• Authentication: Bearer token required on every request. No token, no data.
• Rate limiting: Enforced per API key via governor. Generous for normal usage, capped to prevent abuse.
• Input validation: Per-chain validation on all parameters. Malformed addresses are rejected before they hit the database.
• Body size limit: 64KB per request. More than enough for any MCP tool call.
• Read-only: The MCP server has a read-only database connection pool. No writes, no mutations, no risk to production data.
• Edge security: Cloudflare WAF sits in front of everything. Bearer token enforcement happens at the edge before requests reach the application.

The server runs as a separate binary from the same Docker image as the Slashr API. It's isolated by design.

What You Can Build With This

The MCP server opens up use cases we haven't built ourselves yet. Some ideas:

• Automated portfolio monitoring: An agent that checks your delegations daily and alerts you when a validator's track record deteriorates
• Delegation strategy tools: Compare validators by reliability before you stake, using real incident data rather than projected APY
• Research pipelines: Aggregate incident data across chains for reports, dashboards, or academic analysis
• Operator dashboards: Pull your own validator's incident and scan history into a custom monitoring setup
• Risk scoring models: Feed incident frequency, severity, and scan results into quantitative risk models for staking protocols or insurance products

If you build something interesting with the Slashr MCP server, we'd like to hear about it -- reach out on Twitter or Telegram.

Get Started

1. Generate an API key at slashr.dev/developers
2. Run the claude mcp add command above
3. Start querying live validator data from your AI tools

The data is real, it's live, and it covers four networks. What you do with it is up to you.

Slashr MCP is part of the Slashr platform, built by NullRabbit. The MCP server will be submitted to the Model Context Protocol server registry for public discovery.