How to integrate Twitch chat commands

Let chat summon AI segments on demand. Wire Twitch commands into the HostFabric queue API with rate limits and moderator approvals.

Total time · 20Difficulty · IntermediateTwitch moderator accessServerless function or bot framework
01 · Connect Twitch + scope permissions
4 min

Inside HostFabric, link your Twitch account and specify which commands are public vs moderator-only.

  • Control → Integrations → Connect Twitch
  • Enable !aiquestion for everyone, !airiff for subs only
  • Create API key labelled “Twitch bot”
02 · Set up listener service
6 min

Use EventSub or your preferred bot library to listen for the allowed commands. Validate cooldown windows before forwarding.

  • Deploy serverless function (Vercel/Cloudflare) or run a local bot
  • Store HOSTFABRIC_API_KEY as env var
  • Build cooldown map keyed by userId + command
03 · Map commands to queue payloads
5 min

Craft prompt templates for each command. Include viewer name, paraphrased chat message, and guardrails.

  • Example payload: { "prompt": "Viewer {displayName} asked: {message}" }
  • Attach metadata such as "source": "twitch-command"
  • Send POST /queue with default panel preset
04 · Moderation + review queue
3 min

Flag suspicious inputs for review. Use queue.pause + resume endpoints or route to a backup prompt if mod approval fails.

  • If message includes banned terms, route to `queue.failed` with reason
  • Expose dashboard for mods to approve/deny pending prompts
  • Emit Discord alert for escalations
05 · Instrument analytics
2 min

Log command usage, acceptance rate, and resulting viewer retention. Iterate prompts based on top-performing commands.

  • Send metrics to DataDog/Amplitude
  • Tag queue jobs with command name
  • Share weekly report in the community content hub
Need sample code? Check the developer resources repo for a ready-to-fork Cloudflare Worker that handles !aiquestion to /queue.