launch-kickoff
Skill0 starsChanged 35 days agoMITView on GitHub
Start a product launch in the GrowthCast release checklist studio. Runs intake from a note, PRD or conversation, scores the tier (T1/T2/T3), and pressure-tests the plan. Use for "Fill brief from note" or "Score tier" requests, or when someone says "we're launching X", "plan a product announcement", or "kick off a launch".
In this repository9 other files it gives its agents.
- changelog-automationSkill
- launch-assetsSkill
- launch-linear-syncSkill
- launch-measureSkill
- launch-messagingSkill
- launch-readinessSkill
- launch-retroSkill
- launch-videoSkill
- product-release-checklistSkill
The file
We split this skill into the rules it gives an agent
5 rules. About 0 are common, 1 is distinctive, and 4 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 14
Read the company context file.
- Distinctive: line 17
Use the user's note, and any PRD, Linear or Notion link it contains.
- Very specific: line 23
Ask the user (AskUserQuestion, at most 4 questions per round) only for what you can't infer and what matters most: T-0 date, rollout at …
- Very specific: line 33
No single DRI, or the DRI is the founder (dri).
- Very specific: line 35
No baseline is possible because the events don't exist yet.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Launch Kickoff
Works on a launch JSON (schema in the product-release-checklist skill). If there's no launch yet, create one:
node <tools>/studio.mjs new "<name>" --t0 <yyyy-mm-dd> (ask for the date if it's unknown), and tell the user to open
the studio (/product-release-checklist:open).
1. Context
Read the company context file. Read studio.mjs list to see other launches near this T-0.
2. Intake → brief
Use the user's note, and any PRD, Linear or Notion link it contains. Fill what you can infer:
description: one plain-language paragraphpersona,job("When I…, I want to…, so I can…"),alternative(what they use today)plans,platforms,regions,pricing,outcomerollout.fullandrollout.note, only if stated
Ask the user (AskUserQuestion, at most 4 questions per round) only for what you can't infer and what matters most:
T-0 date, rollout at 100% on T-0, plan eligibility and pricing, and customer evidence. Never invent evidence.
3. Tier → tier.scores, tier.reasons
Score each factor 0–3 using the rubric in studio/model.mjs (RUBRIC). Each reason is one line that cites a fact.
11–15 is T1, 6–10 is T2, 0–5 is T3. Don't set override; the user does that in the studio, with a reason.
4. Pressure-test → feedback
Replace feedback with findings for every failure mode that applies (severity, area, text, fix, owner role, item id):
- The announcement comes before 100% rollout for the announced audience (
rollout-plan). Blocker. - No single DRI, or the DRI is the founder (
dri). - The success metric is a vanity metric rather than activation (
metric). - No baseline is possible because the events don't exist yet. Instrumentation is the critical path (
events). - A T1 with no customer evidence or proof point (
evidence). Blocker for T1. - Pricing or packaging undecided under 14 days out (
packaging). - The feature name is internal jargon (
name). - Another T1/T2 within ±7 days (
calendar). - Company-specific risks from the context file. For example: a feature gated to paid plans needs an upgrade path for Free users; supported languages need localized copy.
Set feedback.at to now and put the single biggest risk in feedback.summary.
5. Hand off
Validate. In chat, give: the tier with a one-line reason, the next 5 items due (from the Checklist tab), and the next step (usually "Write message house").
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

