ICMD AI CODING ASSISTANTS ROLLOUT PLAYBOOK (TEAM CHECKLIST) Use this checklist to roll out AI coding assistants (Cursor, GitHub Copilot, Replit, or similar) without trading speed for reliability. 1) Define your goals (pick 2–3) - Reduce cycle time (idea → merged PR) - Increase test coverage in high-risk services - Reduce on-call incidents while maintaining shipping pace - Shorten onboarding time for new engineers 2) Choose the initial scope (start narrow) - Pick 1–2 teams and 1–2 repos (don’t start with your most critical system) - Pick 2–3 use cases: test drafting, refactors, scaffolding, docs, debugging - Set a pilot window and a decision date 3) Establish access and identity controls - Require SSO for paid tools; avoid personal accounts for company code - Document who can use what (employees, contractors, interns) - Decide whether usage logs are retained and who can audit them 4) Set data handling rules - Ban secrets in prompts (API keys, tokens, private certs) - Decide what proprietary code/context can be sent to third-party inference - Create a red-list of modules (payments, auth, PII) with stricter rules 5) Standardize the prompting format (make intent explicit) Use a shared template: - Context: language, framework, repo conventions - Constraints: “no new deps,” “preserve API,” latency budget, security rules - Acceptance criteria: tests, expected outputs, edge cases - Output format: plan-first, list of files changed, or patch/diff style 6) Update engineering standards for AI-generated code - AI-assisted code must pass the same CI gates (lint, unit tests, type checks) - Require readable diffs: prefer small PRs; avoid giant “AI dumps” - Add a PR note: what was AI-assisted and how it was verified 7) Harden your review process - Create a reviewer checklist for sensitive changes: * Auth/session changes: threat model + negative tests * Payments/billing: idempotency, retries, audit logs * Data migrations: backfill plan, rollback plan, metrics - Require domain-owner review for high-risk modules 8) Add automated guardrails - Enable secret scanning and dependency scanning - Use SAST tooling (for example, Semgrep, Snyk) and enforce severity gates where appropriate - Add pre-commit hooks for formatting and basic checks to reduce review noise 9) Train the team (keep it short and practical) Teach three patterns: (a) ask for a plan before code (b) generate in small chunks (c) bind output to acceptance tests and CI Also share a small set of “golden prompts” drawn from your own codebase. 10) Measure outcomes and decide on expansion Track during the pilot: - Cycle time: time from first commit to merge - Review load: review rounds, PR size distribution, comment density - Quality: bug rate, incident count, severity of pages - Adoption: active users and which use cases actually helped Expansion rule of thumb: - If speed improves without a quality drop, expand to more repos. - If quality drops, tighten constraints: smaller diffs, stronger tests, stricter review. End state: AI becomes a normal co-author, and your process (tests + review + scanning) is the governor that keeps speed and safety in balance.