How to use the Production Launch Checklist agent skill

What the shipping-and-launch skill makes the agent do

An agent skill is a folder containing a SKILL.md file with instructions that the agent reads when a request matches the skill's description. This one is for getting a change ready for production. Its stated goal is to deploy safely, with monitoring in place, a rollback plan ready, and a clear idea of what success looks like. Every launch should be reversible, observable, and incremental.

• A pre-launch checklist covering code quality, security, performance, accessibility, infrastructure and documentation.
• Feature flag guidance: deploy with the flag off, enable for the team, roll out gradually, monitor, then clean up.
• A staged rollout sequence: staging, production with the flag off, the team, a 5% canary, then 25%, 50% and 100%.
• Rollout decision thresholds for error rate, P95 latency, client JS errors and business metrics, saying when to advance, hold and investigate, or roll back.
• A monitoring list for application, infrastructure and client metrics, plus error reporting examples and a first-hour post-launch verification list.

How to install the shipping-and-launch skill

Download the ZIP for this skill from this page and unzip it. The skill is a single SKILL.md file, so you can also place it by hand. Choose one of the locations below depending on whether you want it for all your projects or one project.

To install it for all of your projects, unzip it into your personal skills folder at ~/.claude/skills/. To install it for one project only, unzip it into the .claude/skills/ folder inside that project. In the Claude apps, you can instead upload the ZIP under Settings → Capabilities → Skills.

• Personal (all projects): ~/.claude/skills/
• Single project: .claude/skills/ inside your project folder
• Claude apps: Settings → Capabilities → Skills, then upload the ZIP

How to trigger the skill with natural requests

You do not need a special command. The agent decides to use the skill when your request matches its description. The description names these situations: preparing to deploy to production, asking what needs to be in place before shipping, needing a pre-launch checklist, setting up monitoring, planning a staged rollout, or needing a rollback strategy.

Plain requests that match this description work well. Name the change and the kind of help you want, so the agent knows which part of the process to focus on.

• "We're about to deploy the new checkout flow to production. What needs to be in place first?"
• "Give me a pre-launch checklist for releasing the task sharing feature."
• "Plan a staged rollout behind a feature flag, with rollback criteria."
• "What should we monitor after this release, and what error rate means we roll back?"

Examples of what you get back

When the skill applies, expect the agent to answer with the relevant part of the process rather than a generic list. For a first release, it will usually walk through the checklist sections. For a rollout question, it will give the sequence and the threshold table. For a monitoring question, it will list what to watch and suggest a first-hour check.

Treat the output as a draft to adapt. The checklist items are written for a web application, so you will mark items as not applicable where they do not fit your change.

What the skill needs and what it does not do

The skill lists no requirements. It does not run scripts and does not use the network. It is guidance and checklists. It will not deploy your application, configure DNS, or check your real metrics. You or your team still perform those steps and read your own dashboards.

The advice assumes a web application. The examples use TypeScript, React and Node-style code, so teams on very different stacks will need to translate them.

• Requires: nothing listed
• Runs scripts: no
• Uses network: no

Tips for getting the most from it

Give the agent your real context: what changed, which users it affects, and what your monitoring already shows. The thresholds are relative to a baseline, so the agent needs to know what your normal error rate and latency are.

Use the feature flag rules as written. Each flag should have an owner and an expiration date, and flags should be cleaned up within 2 weeks of full rollout. Avoid nesting flags.

• Name the baseline numbers you have, so the thresholds become concrete.
• Ask for the checklist section by section if the full list is too long for one review.
• Keep the first-hour verification list next to your deploy runbook.