rooster

rooster is a maintenance slack app that performs various duties in the fern slack environment.

open customer threads

if a support thread in #customer-alerts has not been marked with ✅ by the end of the day, it will be included in a round-up of messages, all sent to the #customer-alerts channel, tagging the deployed engineer on-call.

setup

environment variables

rooster requires the following environment variables (see .env.example):

variable	required	description
SLACK_BOT_TOKEN	yes	slack bot token (xoxb-...)
SLACK_SIGNING_SECRET	yes	slack signing secret
SLACK_APP_TOKEN	yes	slack app-level token (xapp-...)
PYLON_API_TOKEN	yes	pylon api token
DEVIN_SLACK_USER_ID	yes	devin's slack user id
DEVIN_TRIAGE_CHANNEL	yes	channel id for #devin-triage-runs
PYLON_WEBHOOK_SECRET	yes	pylon webhook HMAC secret
WEBHOOK_PORT	no	webhook server port (default: 3000)
SLASH_COMMAND	no	slash command name (default: /rooster)

adding a new environment variable

1. add it to .env.example with a placeholder value
2. add it to src/config.ts (use requireEnv if required)
3. add it as a repository secret in GitHub: Settings > Secrets and variables > Actions > New repository secret
4. add it to the .env file on the EC2 instance:

   ssh -i /path/to/your-key.pem <EC2_USER>@<EC2_HOST>
   cd ~/rooster
   nano .env
   # add the new variable, save, then restart:
   docker restart rooster

run the app

pnpm install
pnpm run build
pnpm start

for development:

pnpm run dev

slack commands

• /rooster status - check if rooster is running

• /rooster check [days] - check for issues (default: 1 day)

  • by default, shows both new (unassigned) and waiting-on-you issues in separate sections
  • e.g. /rooster check 3 checks the last 3 days
  • add --new to show only new (unassigned) issues
  • add --open to show only waiting-on-you issues
  • add --channel to post results to #customer-alerts
  • add --remind to tag on-call engineers (implies --channel)

• automatic triage via Pylon webhooks

  • when a new issue is created in Pylon, a webhook fires to rooster's /pylon/webhook endpoint
  • rooster posts the issue context to #devin-triage-runs, @mentioning Devin with triage instructions and on-call handles
  • Devin triages and tags the appropriate on-call team
  • if Slack thread info is available in the payload, rooster also replies in the original thread with a link to the triage run
  • requires PYLON_WEBHOOK_SECRET env var for HMAC signature verification

more functions coming soon...

• triage can decide a thread is incident-worthy (integrate with incident.io, open an incident)
• triage can check account subscription status (connect to supabase)
• alert for stale PRs
• alert for tagged, open issues
• alert immediately for high-risk messages (e.g. "broken", "down", "inaccessible")