agentsclimarketplace

Launch review

Skill Amey-Thakur/AI-SKILLS/skills/big-tech-processes/launch-review

Run a launch review that gates a release behind a readiness checklist, named sign-offs, and a clear no-go authority. Use when a user-facing launch is about to ship and a bad release would be costly to walk back.From its SKILL.md

Install
npx -y skills add Amey-Thakur/AI-SKILLS --skill launch-review

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • 25 days oldThe repository was created 25 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
  • 4 stars4 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

2.8 KB, 589 tokens by cl100k_base, as published. Nobody here has run it

Launch review

A launch review is the last gate between a build and its users: a structured check that the thing is truly ready and that named people have said so on the record. It rots into theater when it becomes a rubber stamp, a meeting where everyone assumes someone else looked and no one holds the authority to stop the train. The whole point is to keep "no" a real and cheap option right up to launch.

Method

  1. Drive the review from a written readiness checklist. Functionality verified, error budget healthy, monitoring and alerts wired, rollback tested, legal and privacy cleared, support and docs ready, comms drafted. A checklist turns "are we ready" from a vibe into items that are each done or not.
  2. Give every checklist item a named owner who signs. Not "engineering approves" but a specific person per area: the on-call lead for operability, privacy counsel for data, the support lead for readiness. A sign-off without a name is a gap nobody owns.
  3. Require the rollback and kill switch to be exercised, not just present. The most common launch-day failure is a revert path that was written and never run. Demand evidence it worked in a realistic environment, with the time-to-revert measured and recorded.
  4. Name one launch owner who holds no-go authority. A single person can halt the launch, and the room knows who it is before the meeting starts. Diffuse authority is how a known-broken launch ships, because stopping it felt like someone else's call.
  5. Set explicit go, no-go, and conditional-go outcomes. Conditional-go names the exact blockers and who confirms they cleared before traffic ramps. "We'll fix it right after launch" is a no-go wearing a go's badge.
  6. Plan the ramp and the watch, not only the flip. Staged rollout by percentage or region, the metrics that define healthy, the threshold that triggers rollback, and who is watching during the window. A launch is not done when it starts; it is done when it is stable.

Checks

  • Does every checklist item carry a named owner who actually signed, not a team label?
  • Has the rollback been executed and timed, or only documented?
  • Is there one person who can say no, and does the room know who it is?

Boundaries

Match the ceremony to the blast radius: a flagged internal rollout does not need the review a public payments launch demands. This skill covers the go decision; the deeper operational readiness of the service underneath belongs to the production-readiness-review skill, and post-launch failures route to a postmortem.

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.

Keep looking

Skills are one crate of 326,679. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.