Cacuda ghost logo CACUDA

Word Lists

3 min read

Overview

Word lists let you define custom sets of words and phrases for content filtering — going beyond ML-based detection to catch specific terms, brand names, slurs, or any domain-specific vocabulary.

Word list checks are free (they don't consume credits), making them an efficient first line of defense in your moderation pipeline.

Creating a Word List

  1. Navigate to /automod/word-lists
  2. Enter a name for your word list (e.g. "Prohibited Terms", "Brand Competitors")
  3. Add words, one per line, in the text area
  4. Select a match mode
  5. Click Save

Match Modes

Word lists support three matching strategies:

Exact

Matches whole word boundaries only. "spam" matches "spam" but not "spammer" or "antispam".

Contains

Matches as a substring anywhere in the text. "spam" matches "spam", "spammer", and "antispam".

Regex

Matches using regular expressions. Use sp[a@]m to match "spam" and "sp@m" (leet speak evasion).

Using in a Flow

  1. Open your automod in the flow builder
  2. Drag a Word List Check processor node from the palette
  3. Click the node and select your word list from the config panel dropdown
  4. Connect the Pass output (no match found) and Fail output (match found) to your next nodes

Combining with ML Detection

A powerful pattern is to use word lists as a free pre-filter before running ML detection:

  1. Trigger receives content
  2. Word list check runs first (free — no credits used)
  3. If a prohibited word is found, route directly to Reject
  4. If no match, pass to an ML detector (uses 1 credit)
  5. Route based on ML score to Approve or Reject
Tip: This pattern saves credits by catching obvious violations with the free word list check before running the paid ML detection.
Next: Flow Builder