Imported from dkeserla/terminal-bench-tests (jobs/2026-04-03__22-58-02/password-recovery__HJNsR8e/agent/.tmp/plugins/plugins/vercel/skills/vercel-firewall/SKILL.md). Install upstream with npx skills add dkeserla/terminal-bench-tests --skill vercel-firewall. Copyright stays with the author.

Vercel Firewall

You are an expert in the Vercel Firewall — a multi-layered security solution with automatic DDoS protection, a customizable Web Application Firewall (WAF), bot management, and rate limiting.

Architecture & Rule Execution Order

1. DDoS mitigation rules (automatic, platform-wide)
2. WAF IP blocking rules
3. WAF custom rules (in priority order)
4. WAF Managed Rulesets (OWASP, Bot Protection, AI Bots)

Changes propagate globally in under 300ms. No redeployment required.

DDoS Protection (Automatic, All Plans)

• Layer 3/4 mitigation (automatic, always on)
• Layer 7 protection (proprietary, tailored to web apps)
• Protectd: Vercel's DoS mitigation infrastructure analyzes ~550K events/sec globally with median mitigation time of 2.5 seconds
• 40x faster detection with real-time stream processing
• Handles 1B+ suspicious TCP connections per week
• Proven to mitigate 1.37 Tbps attacks with zero downtime

No configuration needed — DDoS protection is always active.

WAF Custom Rules

Rule JSON Structure

{
  "name": "Block WordPress scanners",
  "description": "Block common WordPress probe paths",
  "active": true,
  "conditionGroup": [
    {
      "conditions": [
        {
          "type": "path",
          "op": "re",
          "value": "^/wp-(admin|login|content|includes)/"
        }
      ]
    }
  ],
  "action": {
    "mitigate": {
      "action": "deny"
    }
  }
}

Logic: Each object in conditionGroup is an OR group. Conditions within a single group are ANDed. Multiple groups are ORed.

Condition Types (25 available)

Condition Operators

Additional optional fields: neg: true negates the condition, key required for header/query/cookie types.

Mitigation Actions

Persistent Actions

By default each request is evaluated individually. With persistent actions, rules are applied to all matching requests for a customizable duration (actionDuration), allowing the firewall to remember malicious behavior and block it earlier in the lifecycle.

Action Options

{
  "action": {
    "mitigate": {
      "action": "deny",
      "actionDuration": "1h",
      "bypassSystem": false,
      "logHeaders": ["user-agent", "x-forwarded-for"],
      "redirect": {
        "location": "https://example.com/blocked",
        "permanent": false
      }
    }
  }
}

Practical Rule Examples

Block Sanctioned Countries

{
  "name": "Block OFAC Sanctioned Countries",
  "active": true,
  "conditionGroup": [
    {
      "conditions": [
        {
          "type": "geo_country",
          "op": "inc",
          "value": ["CU", "IR", "KP", "RU", "SY"]
        }
      ]
    }
  ],
  "action": {
    "mitigate": { "action": "deny" }
  }
}

Require API Key Header on /api/ Routes

{
  "name": "Require API Key",
  "active": true,
  "conditionGroup": [
    {
      "conditions": [
        {
          "type": "header",
          "op": "nex",
          "key": "x-api-key"
        },
        {
          "type": "path",
          "op": "pre",
          "value": "/api/"
        }
      ]
    }
  ],
  "action": {
    "mitigate": { "action": "deny" }
  }
}

Block by JA4 TLS Fingerprint

{
  "name": "Block Known Malicious JA4",
  "active": true,
  "conditionGroup": [
    {
      "conditions": [
        {
          "type": "ja4_digest",
          "op": "eq",
          "value": "t13d1516h2_8daaf6152771_b0da82dd1658"
        }
      ]
    }
  ],
  "action": {
    "mitigate": { "action": "deny", "actionDuration": "1h" }
  }
}

Block Datacenter ASNs

{
  "name": "Block Known Datacenter ASNs",
  "active": true,
  "conditionGroup": [
    {
      "conditions": [
        {
          "type": "geo_as_number",
          "op": "inc",
          "value": ["14618", "16509", "15169"]
        }
      ]
    }
  ],
  "action": {
    "mitigate": { "action": "deny" }
  }
}

Challenge cURL Requests

{
  "name": "Challenge cURL",
  "active": true,
  "conditionGroup": [
    {
      "conditions": [
        { "type": "user_agent", "op": "re", "value": "^curl/" }
      ]
    }
  ],
  "action": {
    "mitigate": { "action": "challenge" }
  }
}

Rate Limiting

Rate Limit Rule

{
  "name": "API Rate Limit - 100 req/min",
  "active": true,
  "conditionGroup": [
    {
      "conditions": [
        { "type": "path", "op": "pre", "value": "/api/" }
      ]
    }
  ],
  "action": {
    "mitigate": {
      "action": "rate_limit",
      "rateLimit": {
        "algo": "fixed_window",
        "window": 60,
        "limit": 100,
        "keys": ["ip"],
        "action": "deny"
      }
    }
  }
}

Login Endpoint Protection

{
  "name": "Login Rate Limit",
  "active": true,
  "conditionGroup": [
    {
      "conditions": [
        { "type": "path", "op": "eq", "value": "/api/auth/login" },
        { "type": "method", "op": "eq", "value": "POST" }
      ]
    }
  ],
  "action": {
    "mitigate": {
      "action": "rate_limit",
      "rateLimit": {
        "algo": "fixed_window",
        "window": 60,
        "limit": 10,
        "keys": ["ip"],
        "action": "challenge"
      }
    }
  }
}

Rate Limit Configuration Options

When exceeded with deny, returns HTTP 429 with X-RateLimit-Limit and X-RateLimit-Remaining headers.

Bot Management

Bot Protection (GA — Free on All Plans)

Heuristics-based detection that challenges non-browser bot traffic without disrupting verified webhook providers. Formerly "Bot Filter" during beta — renamed to Bot Protection at GA. Enable in log-only mode first to preview traffic impact:

{
  "action": "managedRules.update",
  "id": "bot_protection",
  "value": { "active": true, "action": "challenge" }
}

Note: The older bot_filter ID is deprecated. Use bot_protection in new configurations.

AI Bot Blocking

Block known AI crawlers (GPTBot, ClaudeBot, etc.):

{
  "action": "managedRules.update",
  "id": "ai_bots",
  "value": { "active": true, "action": "deny" }
}

Allow a Specific Bot (Bypass Rule)

Place this higher in priority than Bot Protection managed rules:

{
  "name": "Allow My Monitoring Bot",
  "active": true,
  "conditionGroup": [
    {
      "conditions": [
        { "type": "user_agent", "op": "eq", "value": "MyMonitorBot/1.0" }
      ]
    }
  ],
  "action": {
    "mitigate": { "action": "bypass" }
  }
}

Enable BotID (Traffic Visibility)

{ "botIdEnabled": true }

IP Allow/Block Lists

Block an IP

{
  "action": "ip.insert",
  "value": {
    "hostname": "my-site.com",
    "ip": "203.0.113.45",
    "action": "deny",
    "notes": "Malicious scraper"
  }
}

Block a CIDR Range

{
  "action": "ip.insert",
  "value": {
    "hostname": "my-site.com",
    "ip": "203.0.113.0/24",
    "action": "deny",
    "notes": "Bad actor CIDR block"
  }
}

Allow an IP (Bypass All Rules)

{
  "action": "ip.insert",
  "value": {
    "hostname": "my-site.com",
    "ip": "198.51.100.1",
    "action": "bypass",
    "notes": "Internal monitoring IP"
  }
}

IP Rule Actions

Note: hostname must match the exact domain. Add separate entries per subdomain.

OWASP Core Ruleset (CRS)

Individual CRS Rules

Enable OWASP Rules

{
  "action": "crs.update",
  "id": "sqli",
  "value": { "active": true, "action": "deny" }
}

Full OWASP + Bot Configuration (PUT)

{
  "firewallEnabled": true,
  "crs": {
    "sqli": { "active": true, "action": "deny" },
    "xss": { "active": true, "action": "deny" },
    "rce": { "active": true, "action": "deny" },
    "lfi": { "active": true, "action": "deny" },
    "rfi": { "active": true, "action": "deny" },
    "sd": { "active": true, "action": "log" },
    "ma": { "active": true, "action": "deny" },
    "gen": { "active": true, "action": "deny" },
    "sf": { "active": true, "action": "deny" },
    "php": { "active": false, "action": "log" },
    "java": { "active": false, "action": "log" }
  },
  "managedRules": {
    "owasp": { "active": true, "action": "deny" },
    "bot_protection": { "active": true, "action": "challenge" },
    "ai_bots": { "active": true, "action": "deny" }
  },
  "botIdEnabled": true
}

Firewall REST API

Base URL: https://api.vercel.com Auth: Authorization: Bearer <VERCEL_TOKEN> Query params: ?projectId=<id>&teamId=<id>

Endpoints

PATCH Actions

Add a Rule via cURL

curl -X PATCH "https://api.vercel.com/v1/security/firewall/config?projectId=prj_xxx&teamId=team_xxx" \
  -H "Authorization: Bearer $VERCEL_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "action": "rules.insert",
    "value": {
      "name": "Block WordPress scanners",
      "active": true,
      "conditionGroup": [
        {
          "conditions": [
            { "type": "path", "op": "re", "value": "^/wp-(admin|login|content|includes)/" }
          ]
        }
      ],
      "action": { "mitigate": { "action": "deny" } }
    }
  }'

Vercel SDK Usage

import { Vercel } from '@vercel/sdk'

const vercel = new Vercel({ bearerToken: process.env.VERCEL_TOKEN })

// Read current firewall config
const config = await vercel.security.readFirewallConfig({
  configVersion: 'active',
  projectId: 'prj_xxx',
  teamId: 'team_xxx',
})

// Add a rule
await vercel.security.updateFirewallConfig({
  projectId: 'prj_xxx',
  teamId: 'team_xxx',
  requestBody: {
    action: 'rules.insert',
    value: {
      name: 'Rate limit API',
      active: true,
      conditionGroup: [
        { conditions: [{ type: 'path', op: 'pre', value: '/api/' }] },
      ],
      action: {
        mitigate: {
          action: 'rate_limit',
          rateLimit: { algo: 'fixed_window', window: 60, limit: 100, keys: ['ip'], action: 'deny' },
        },
      },
    },
  },
})

Create Temporary Bypass (Attack Challenge Mode)

curl -X POST "https://api.vercel.com/v1/security/firewall/bypass?projectId=prj_xxx&teamId=team_xxx" \
  -H "Authorization: Bearer $VERCEL_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "domain": "my-site.com",
    "sourceIp": "198.51.100.42",
    "ttl": 3600000,
    "note": "Temporary bypass for load testing"
  }'

vercel.json WAF Rules

Declaratively define firewall rules in vercel.json using the mitigate key:

{
  "$schema": "https://openapi.vercel.sh/vercel.json",
  "routes": [
    {
      "src": "/api/(.*)",
      "missing": [
        { "type": "header", "key": "x-internal-token" }
      ],
      "mitigate": { "action": "deny" }
    },
    {
      "src": "/(.*)",
      "has": [
        { "type": "header", "key": "user-agent", "value": "(?i)^curl/" }
      ],
      "mitigate": { "action": "challenge" }
    }
  ]
}

Supported actions in vercel.json: "challenge", "deny" only. Rate limiting, log, and bypass require the Vercel Firewall dashboard at https://vercel.com/{team}/{project}/firewall or the REST API.

Attack Challenge Mode

• Available on all plans (free)
• Shows browser verification challenge to all visitors during active attacks
• Legitimate bots (Googlebot, webhook providers) automatically pass through
• Internal Function-to-Function calls within the same account bypass automatically
• Blocked requests don't count toward CDN/traffic usage
• Configured via dashboard only: open https://vercel.com/{team}/{project}/firewall → Bot Management → Attack Challenge Mode

Plan Availability

Observability

• Security event logs in the Firewall tab
• IP enrichment — hover any IP in the Firewall dashboard to see ASN, location, and metadata
• Create custom WAF rules directly from dashboard traffic charts (select "Create Custom Rule" from the actions menu)
• Linkable to Monitoring queries for investigations
• DDoS mitigation notifications (alerts on detection)
• BotID traffic visibility when enabled

Official Documentation

• Vercel Firewall Overview
• Custom Rules
• Rate Limiting
• IP Blocking
• Managed Rulesets
• Attack Challenge Mode
• Firewall API Guide
• REST API Reference