AI Team Norms Template

Team AI Norms Working Agreement
Version: [date] | Review cadence: [sprint frequency]

1. Disclosure
   We [do/do not] indicate AI tool contribution in PR descriptions.
   Minimum disclosure: "Significant AI contribution to [file/function]"

2. Review Standards
   All code, regardless of origin, receives a review by someone who did not write it.
   For AI-heavy PRs, the reviewer confirms they read and understood the logic.

3. Conservative Use Cases
   We apply extra scrutiny to AI-generated code in: [authentication / payment flows / safety logic / other]

4. Failure Disclosure Norm
   When AI tools produce something incorrect, we share it with the team. This is an achievement, not an embarrassment.

5. Disagreement Protocol
   When team members disagree about whether AI-generated code meets our standards, we [discuss in PR / escalate to retrospective / require second review]

6. Review Schedule
   We will revisit this agreement in retrospective every [N] sprints.

// Actual team agreement after facilitated working session:

norms: {
  disclosure: "PR description includes: AI-assisted (significant) when AI wrote >50% of changed lines",
  reviewStandard: "Reviewer must be able to explain the logic to a new team member — not just confirm tests pass",
  conservativeAreas: ["auth middleware", "payment processing", "user data handling"],
  failureDisclosure: "Standup acknowledgment of AI failures celebrated — Scrum Master explicitly thanks the person",
  disagreementProtocol: "Second review required; if still unresolved, 15-minute synchronous conversation before merge",
  reviewCadence: "Every 3 sprints in retrospective"
}

facilitationNote: "This took 45 minutes in a dedicated working session. Trying to do it in a retrospective sidebar produced a vague agreement nobody referred to."