More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Shipping and Launch: Production Readiness

    1

    Launch to production with a readiness review, flag-gated staged rollout, numeric advance/hold/rollback thresholds agreed in advance, and a rehearsed rollback.

    $5

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Shipping and Launch: Production Readiness

    Shipping and Launch: Production Readiness

    Example session with this skill installed

    We're launching a redesigned onboarding flow next week for our SaaS app, about 4,000 active users. It includes a database migration (new table, backfill from the old one). We have feature flags and basic error/latency dashboards. Give us the launch plan: readiness, rollout steps, and when we'd roll back.

    • Read your context and instructions
    • Compiled the shipping and launch

    Launch plan — onboarding redesign + migration.

    Readiness verdicts. Correctness: ready when the suite is green and both flag states are tested in CI — the off path rots silently otherwise. Data: GAP until proven otherwise. Your backfill must be timed on a production-shaped snapshot and tested down as well as up; a 40-minute surprise migration is a self-inflicted incident. Rewrite it as a batched backfill if it exceeds a few minutes. Write down what happens to rows the new flow writes if you roll back — if old code can't read them, rollback stops being a flag flip and the plan changes. Security: ready after a dependency audit and a check that no new endpoints ship unauthenticated. Performance: measure the new flow's completion time against the old one; onboarding is your funnel, so "feels fine" isn't a number. Operability: dashboards exist; alert thresholds are untested — accept as a named gap with an owner, or test them this week. Comms: changelog drafted; support gets a one-paragraph heads-up before the canary, not after the first ticket.

    Rollout ladder. Monday: deploy dark, flag off; verify health and error rate. Enable for the team on real accounts. Tuesday: canary at 5% for at least four business hours. Then 25% → 50% → 100% on separate days, pausing at each rung until signup-completion numbers mean something.

    Thresholds, agreed now. Advance: error rate and latency within a small margin of baseline, no new error types, signup completion neutral or better. Hold: anything elevated but ambiguous — latency up ~30%, a new error at tiny volume; investigate at that rung, never advance on a mystery. Roll back: error rate doubles, signup completion drops decisively, any data-integrity question, or a security issue. Rollback mechanism is the flag flip (seconds); the additive migration can stay in place harmlessly with the flag off.

    First hour of each rung: a named watcher on the dashboards, the signup flow exercised by hand, logs checked for new warnings. Roles named in advance: launch lead (sole owner of the advance/hold/rollback call), watcher, comms owner posting every 30 minutes, rollback hand. Flag expiry: two weeks after 100%; retro scheduled.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Most launch disasters aren't code failures — they're launches with no agreed thresholds, nobody watching the first hour, and a rollback plan being invented mid-incident. This skill builds the launch as a controlled operation. It starts with a readiness review across the six dimensions that actually sink launches (correctness, data, security, performance, operability, comms), each scored ready / gap / blocker with evidence, not vibes. Then it decouples deploy from release: dark deploy behind a feature flag, team enablement, canary, and a 25/50/100 ramp. Every rung gets pre-agreed numeric thresholds — advance, hold-and-investigate, or roll back — written down while everyone is calm. It includes a launch-plan template, named launch roles and a comms cadence, the first-hour watch checklist, a written rollback plan covering triggers, mechanism, data handling, and communication, and the post-launch discipline: flag cleanup, a full business cycle of monitoring, and a retro that feeds the next launch. A complete worked example shows a checkout launch where the hold threshold caught a real problem at 25% and the system worked as designed.

    • It writes and structures the plan; your team still executes the launch, watches the first hour, and rehearses the rollback.
    • Thresholds are set from your baselines — if you have no baseline error/latency numbers yet, recording them becomes a readiness gap the plan will flag.
    • Store-shipped mobile and desktop releases can't be rolled back once distributed; the skill covers the compensating controls (kill switches, staged store rollouts) but can't change that constraint.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 24 days ago

    Permissions

    No special permissions declared or detected

    Listed24 days ago
    Updated21 days ago

    What's inside

    Frequently Asked Questions

    Yes — the ladder matters more than the tooling. Use the substitutes the skill lays out: deploy to a subset of servers or a percentage of traffic at the load balancer, or gate the new behavior by account allowlist behind a config switch. What you can't skip is the staging itself and thresholds agreed before you start.

    Scale it to the risk, keep the skeleton: a one-paragraph readiness check, deploy dark or canary-small, one numeric rollback trigger, a named watcher for the first hour. The full ceremony is for launches where failure hurts users or data. A change that can't hurt either doesn't need this skill at all.

    That's exactly the failure the thresholds and roles exist for. The rollback triggers are numbers agreed before launch day, and one named launch lead holds the call — so rolling back becomes executing the plan, not admitting defeat in front of stakeholders. Teams that pre-agree the trigger almost never argue about it in the moment.

    New in Business & Operations