DROP LIVE · Claude Shipper v3 · 64/100 claimedNEW · Editorial UI Pack · 22/60 claimedSOON · Drop Launch Playbook · opens FriVault XP × 2 this week187 prompts shipped in VaultFree guide · How I shipped a project in 47 minutesDROP LIVE · Claude Shipper v3 · 64/100 claimedNEW · Editorial UI Pack · 22/60 claimedSOON · Drop Launch Playbook · opens FriVault XP × 2 this week187 prompts shipped in VaultFree guide · How I shipped a project in 47 minutes
← back to drops
MVP
LIVE#skills
4d left
#skillsclaudecursoragents
K3 · ED. 2026
curator notes · Mira O.

Why this kit, and why now.

What this kit gives you

A working agent loop that ships small features in under an hour. Six months of refining a single workflow into the precise stack that turned a five-day rebuild into one Saturday afternoon. Includes the kill-switch rule for runaway agents, the prompt sequence the agent runs in, and the post-mortem playbook that catches the bugs the agent did not.

This is the room you want to be in if you have started writing your tenth "I will just have Claude do it" project.

Why this works

Most agent kits ship prompts. Prompts decay. Models change, tasks change, your taste changes, and last week's prompt is this week's noise.

What survives is a small, opinionated rules layer that constrains the agent to one decision pattern: plan before you do, ask before you change a public API, write the post-mortem before you ship. This kit codifies that pattern in three Cursor rules and one Claude skill, plus a runbook the agent reads first on every loop.

Who this is for

  • You ship in Cursor and you want the agent to stop creating drift between your plan and the code that lands.
  • You build with Claude (Pro or Max) and you want a workflow that survives model upgrades.
  • You have shipped one or two real projects with AI assist and you want the next one to feel deliberate.

If you are still finding your way into AI-assisted shipping, start with the free companion guide first. The guide walks the timeline of a real 47-minute ship. The kit gives you the rules and the runbook to repeat it.

What you get

Four files in a tiny zip:

  • claude-shipper.skill.json — the Claude skill that wires the agent loop
  • cursor-rules.mdc — three rules the agent reads on every prompt
  • playbook.pdf — the post-mortem checklist (printable)
  • 47min-runlog.md — annotated log of the original 47-minute build, prompt by prompt

You own the kit. No DRM, no expiry, lifetime updates when Claude or Cursor breaks the workflow.

How to use it

  1. Drop the rules file into your project root as .cursorrules.
  2. Install the Claude skill via the kit README (one CLI command).
  3. Open the agent in Cursor and run the runbook prompt.
  4. Stop reading and start shipping.

Companion guide

How I shipped a client project in 47 minutes — free read, real timeline.

WHAT YOU GET4 files
SK
claude-shipper.skill.json
skill
12 KB
RU
cursor-rules.mdc
rules
4 KB
PD
playbook.pdf
pdf
1.2 MB
LO
47min-runlog.md
log
9 KB
Cut a 3-day refactor to a coffee break. The kill-switch rule alone is worth the price.
Aiden K., Solo founder, Remix.dev
verified buyer
Finally a kit that respects my time. No fluff, all signal.
Sara P., Indie agent dev
verified buyer
FOUNDATION GUIDE · FREE

How I shipped a client project in 47 minutes

A timeline, the prompts, the kill switch, and what almost broke.

read first →
CURATOR
Mira O.
Ships kits that respect your Saturday.
EDITION
100 copies · 4d window
After Friday: archived to alumni-only.
PAIRS WITH

Other drops on the same shelf