Plan a feature flag rollout and rollback
Designs a staged release behind a feature flag, with audience steps, metrics to watch, kill criteria, and flag cleanup.
The prompt
Help me plan the rollout of a feature behind a feature flag.
Feature: [WHAT CHANGES FOR USERS OR SYSTEMS]
Flag tooling: [e.g. "LaunchDarkly", "Unleash", "homegrown config table", "env variable"]
Blast radius if it breaks: [e.g. "checkout fails", "slower search", "cosmetic only"]
Who must be told: [SUPPORT, SALES, OTHER TEAMS, CUSTOMERS]
Metrics I can watch: [ERROR RATES, LATENCY, CONVERSION, BUSINESS KPIs - or "not sure"]
Build the plan:
1. Flag design: flag name, default value, what the "off" path does, and whether old and new code paths can safely flip back and forth (call out data written by the new path that the old path can't read).
2. Rollout stages, e.g. internal staff, 1%, 10%, 50%, 100%, with how long to hold at each and how users are bucketed (sticky by user, account, or region).
3. Go/no-go criteria per stage: the specific metrics and thresholds to check. Use [SET THRESHOLD] where I haven't given baselines.
4. Kill switch: who can turn it off, how fast it takes effect, and what happens to in-flight requests.
5. Communication: who to notify at each stage, plus a 2-line message for support.
6. Cleanup: when to remove the flag and the old code path, with a reminder ticket.
Return the plan as a table of stages, followed by the risks list and the cleanup checklist.
Highlighted [brackets] are blanks to replace with your own details.
Fill in the blanks and use this promptMore Software Development prompts
What to fill in
- WHAT CHANGES FOR USERS OR SYSTEMS
- e.g. "checkout fails", "slower search", "cosmetic only"
- SUPPORT, SALES, OTHER TEAMS, CUSTOMERS
- SET THRESHOLD
How this prompt is built
This prompt follows the Few-Shot structure (Few-shot prompting). A few-shot prompt includes one or more worked examples of the input and the output you want, so the model can copy the pattern instead of guessing it.
Related prompts
- Plan cross-browser and cross-device compatibility testingA prioritized browser and device test matrix, tooling choices, and checklist so your site or app works for your real users.
- Add useful logging, metrics, and tracing to a servicePlans the instrumentation a service needs so you can answer what broke, where, and for whom when something goes wrong.
- Respond to a secret leaked in a code repositoryWalks you through containing a leaked API key, password, or token: revoke, rotate, check for misuse, and clean up history.
- Implement a feature from a spec, with testsPlans and writes production-quality code for a feature, including tests and edge cases.