Walk a user through setup or installation one step at a time with the remaining steps visible.
SkillScore64Maintained
On-demandOn-demand: invoked explicitly — never occupies your agent's context until called.
◆Support pack
Last verified 8/11/2026
What it changes
Passed 14 of 22 test cases with the skill, and 9 of 22 without it.
Without skill41%
With skill64%
Checks it passed that the model alone missed
The first step specifies importing the official Elastic GPG repository key.
The current step isolates checking WSL2 prerequisites into a single atomic action.
The remaining steps list includes running `npm run dev` to start the development server.
and 2 more in the full report
5 fixed9 passed either way8 still failing0 broke
Pass-rate lift
+23 pp
Tokens
-4%
Agent turns
0%
Response time
-44%
Model gemini-3.6-flashMeasured 8/11/2026v114 of 22 cases passed with skill
Tokens, turns and time describe resource use, not a dollar price.
See it in action
One of 5 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 5
Prompt
Please assist me in setting up a new Next.js project with Tailwind CSS and TypeScript from scratch. Give me instructions one action at a time. Format your message with the active step first, put a divider line under it, and then list the remaining steps below so I know what's left to do.
What was checked, without and with the skill
Checked
Without skill
With skill
The first step isolates running create-next-app into one atomic action.
Passed
Passed
The remaining steps list includes running `npm run dev` to start the development server.
Missed
Passed
case-03 · gemini-3.6-flash · 8/11/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md33 linessingle file
setup-help
When to Use
Use when the user asks to set up, install, configure, or get something working step by step.
Use when the setup has multiple steps and benefits from one-at-a-time guidance.
Guide the user through any setup, one step at a time, in plain English.
Response format (every single response)
Current step — ONE atomic action. A single click, field, or command — not a checklist. 1–2 lines max. If it needs sub-steps, it's too big: split it and push the rest into "Still remaining". Plain English.
A ---- divider.
Still remaining — a numbered list of the setup steps left after this one. Max 8 items, ever.
Repeat this format for every response until setup is done.
Rules
Before the first step, build a complete canonical checklist from the user's outline, repo/docs, current screen, and any discovered prerequisites.
The Still remaining list must never exceed 8 items — more is overwhelming. Track ALL unfinished checklist items internally; if more than 8 remain, show the nearest steps individually and merge the later ones into broader phase-level items so the list stays at 8 or fewer. Never silently drop a required step from internal tracking.
If a new required step is discovered mid-setup, add it to Still remaining immediately in the correct order.
Before every response, audit the current step plus Still remaining against the canonical checklist. If any unfinished step is missing, fix the list before replying.
Only give instructions for the current step. Do not jump ahead.
Keep it concise. Short sentences. No filler.
After the user finishes a step, move the next "remaining" item up to "Current step".
Update the "Still remaining" list each time as steps get done.
When nothing remains, say setup is complete instead of showing the list.
Limitations
Adapted from davidondrej/skills; verify local paths, tools, credentials, and agent features before acting.
For commands, remote access, scheduling, browser automation, or file-changing workflows, get explicit user approval and confirm the target environment first.