krawl.es/config.yaml

# Krawl Honeypot Configuration

server:
  port: 5000
  delay: 100  # Response delay in milliseconds

  # manually set the server header, if null a random one will be used.
  server_header: null

links:
  min_length: 5
  max_length: 15
  min_per_page: 5
  max_per_page: 10
  char_space: "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789"
  max_counter: 10

canary:
  token_url: null  # Optional canary token URL
  token_tries: 10

dashboard:
  # if set to "null" this will Auto-generates random path if not set
  # can be set to "/dashboard" or similar <-- note this MUST include a forward slash
  # secret_path: super-secret-dashboard-path
  secret_path: null

backups:
  path: "backups"
  cron: "*/30 * * * *"
  enabled: false

exports:
  path: "exports"

logging:
  level: "DEBUG"  # DEBUG, INFO, WARNING, ERROR, CRITICAL

database:
  path: "data/krawl.db"
  retention_days: 30

behavior:
  probability_error_codes: 0  # 0-100 percentage

analyzer:
  http_risky_methods_threshold: 0.1
  violated_robots_threshold: 0.1
  uneven_request_timing_threshold: 0.5
  uneven_request_timing_time_window_seconds: 300
  user_agents_used_threshold: 2
  attack_urls_threshold: 1

crawl:
  infinite_pages_for_malicious: true
  max_pages_limit: 250
  ban_duration_seconds: 600