Blackwall MCP server

blackwall-mcp

A guardrail for AI agents, as an MCP server. Your agent calls one tool — forecast — before any irreversible action (send email, move money, run SQL, delete data, post content). It gets back a risk score (0–100), a reversibility class, a GO / CAUTION / STOP recommendation, and named red flags in a few seconds (~4-8s).

Works in any MCP host: Claude Desktop, Claude Code, Cursor, Windsurf, and any agent framework with MCP support.

The wall between your agent and disaster. A BLUETIER product.

1. Get an API key

Sign up free at https://blackwalltier.com → Dashboard → API keys → Create key. Free tier: ~100 forecasts/month, no card. Your key looks like bwlive….

2. Add the server to your MCP host

Claude Desktop

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

{
  "mcpServers": {
    "blackwall": {
      "command": "npx",
      "args": ["-y", "blackwall-mcp"],
      "env": { "BLACKWALL_API_KEY": "bw_live_your_key_here" }
    }
  }
}

Restart Claude Desktop. You'll see a forecast tool available.

Cursor

Settings → MCP → Add new global MCP server, then in mcp.json:

{
  "mcpServers": {
    "blackwall": {
      "command": "npx",
      "args": ["-y", "blackwall-mcp"],
      "env": { "BLACKWALL_API_KEY": "bw_live_your_key_here" }
    }
  }
}

Claude Code

claude mcp add blackwall -e BLACKWALL_API_KEY=bw_live_your_key_here -- npx -y blackwall-mcp

Run locally (any host / testing)

BLACKWALL_API_KEY=bw_live_your_key_here npx -y blackwall-mcp

3. Use it

Once added, instruct your agent: "Before any irreversible action, call the forecast tool and stop if it returns STOP." The model will call it automatically when it's about to do something risky.

The forecast tool

Returns: recommendation (GO/CAUTION/STOP), riskscore (0–100), reversibility (class + rollback cost), gate (proceed/confirm/human-required), confidence, redflags[], predictedresult, alternativeactions[].

Example

Agent about to run DELETE FROM users; (no WHERE clause) →

🛑 BLACK_WALL: STOP — risk 99/100
Red flags:
  • [CRITICAL] SQL_NO_WHERE — deletes the entire table, not one row
  • [CRITICAL] INTENT_MISMATCH — intent was "remove a single test row"
  • [CRITICAL] IRREVERSIBLE_NO_BACKUP — no recovery path
Guidance: DO NOT take this action. Surface the red flags to the user.

Observe mode — try it with zero risk

Not ready to let a guardrail block your agents? Start in observe mode. It scores and logs every action but never tells the agent to stop — your agents behave exactly as they do today. After a week, review your dashboard and see what it would have caught.

{
  "mcpServers": {
    "blackwall": {
      "command": "npx",
      "args": ["-y", "blackwall-mcp"],
      "env": {
        "BLACKWALL_API_KEY": "bw_live_your_key_here",
        "BLACKWALL_MODE": "observe"
      }
    }
  }
}

Then see "what your agents almost did" in your dashboard. Flip BLACKWALL_MODE to enforce (or just remove it — enforce is the default) when you're ready to actually block.

Two tools

The server exposes two MCP tools:

Shortened. The full README is on GitHub.