Slack DevOps Onboarding Bot Prompt
Design a Slack onboarding bot for new DevOps / SRE hires — Day-1 checklist, account provisioning verification, runbook tour, on-call shadow scheduling, and 30/60/90-day check-ins.
- Target user
- Engineering managers reducing onboarding time and ramp friction for SRE hires
- Difficulty
- Intermediate
- Tools
- Claude, ChatGPT
The prompt
You are a senior engineering manager who has built Slack-driven onboarding for SRE hires that cut their time-to-first-on-call from 12 weeks to 6 weeks with measurable confidence.
I will provide:
- Team size + on-call rotation cadence
- Tools the team uses (Kubernetes, Terraform, Datadog, PagerDuty, etc.)
- Existing onboarding artifacts (docs, runbooks, learning paths)
- SSO + provisioning automation in place
Your job:
1. **Day 1 sequence** — bot DMs the new hire on first login:
- Welcome message with manager + buddy named
- Account provisioning check (does Okta show all groups assigned?)
- "Reply DONE when:" checklist:
- Cloned the main monorepo
- Joined the team channel
- Set up local dev env (link to setup guide)
- Read the team's "How we operate" doc
- Took the laptop-hardening photo for IT
2. **Week 1 sequence** — daily nudge per workday:
- Day 2: meet 3 services — bot links to one-pager per service + asks new hire to summarize back
- Day 3: tools tour — Datadog dashboards, Grafana, PagerDuty schedule, ArgoCD UI
- Day 4: read 3 recent postmortems; reply with one question per
- Day 5: shadow first on-call session via Zoom; bot schedules it
3. **Weeks 2-4 sequence** — increasing autonomy:
- Week 2: pair on a small infra change PR
- Week 3: handle one low-severity alert with buddy backup
- Week 4: present the team's architecture to the bot (recorded for buddy review)
4. **Knowledge check-ins** — bot quizzes the new hire on key facts:
- "What does our SLO target on the auth service mean in real terms?"
- "Where do you go to roll back a deployment in service X?"
- "Who's the secondary on-call this week?"
- Multiple choice + free text; results route to manager + buddy for follow-up
5. **Runbook tour** — bot walks through 5 most-run runbooks:
- Shows the runbook
- Asks the new hire to articulate when they'd use it
- Records their answer for buddy review
6. **30/60/90-day check-ins**:
- 30d: how confident are you on (services / tools / on-call)? Bot asks via 5-point Likert with comment field
- 60d: what's confusing? What docs are wrong/missing?
- 90d: ready for primary on-call? Manager + buddy review the answer
7. **Bot-to-human handoff** — if the new hire is stuck:
- Replies like "I don't know what X means" → bot routes to buddy
- No reply in 48h → manager pinged
- Low confidence on 30d check → manager + buddy convene
8. **Cohort comparison** — for managers, aggregated dashboard:
- Where do new hires consistently get stuck?
- Which runbooks need better docs (high stuck-rate)?
- Average time-to-first-solo-on-call
9. **Data minimization** — bot stores: completion state, question responses (for manager review), confidence scores. Bot does NOT store: keystroke logs, private chats, personal opinions on coworkers.
10. **Anti-patterns to avoid** — bot as surveillance, bot replaces 1:1 human contact, bot quizzes that don't track to real-world readiness, "completion" that doesn't translate to actual capability.
Output as: (a) Day 1 / Week 1 / Weeks 2-4 message scripts, (b) knowledge-check question bank, (c) runbook tour structure, (d) 30/60/90 questionnaire, (e) bot-to-human escalation rules, (f) manager dashboard schema, (g) data retention + privacy policy.
Bias toward: human contact + bot scaffolding, measurable outcomes, killing the bot's prompts once the hire's ramped (not perpetual nagging).