Agent Yolo Build banner
timkosters timkosters

Agent Yolo Build

AI community

Description

Run /yolo, walk away, come back to a working MVP. Autonomous build skill for AI coding agents.

Installation

This entry records only its repository, not the path inside it, so there is no exact command to give. Open the source below and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

README

/yolo — Autonomous Agent Sessions

Run `/yolo`, walk away, come back to results. Your AI coding agent picks a goal and works on it while you're gone.

Inspired by [the overnight build pattern](https://x.com/bobbyhansenjr/status/2017585853418152313).

Modes

Command What the agent does
/yolo or /yolo build Finds an opportunity and builds a working MVP from scratch
/yolo refactor Finds the highest-leverage improvement in your codebase
/yolo research [topic] Deep research with sources, not just a summary
/yolo explore Reads your notes and projects, surfaces connections you missed

You can also pass a focus: `/yolo build chrome extension` or `/yolo research state of MCP servers`.

Install

Claude Code

mkdir -p ~/.claude/skills/yolo
curl -o ~/.claude/skills/yolo/SKILL.md \
  https://raw.githubusercontent.com/timkosters/agent-yolo-build/main/SKILL.md

Or clone and symlink:

git clone https://github.com/timkosters/agent-yolo-build.git
mkdir -p ~/.claude/skills
ln -s "$(pwd)/agent-yolo-build" ~/.claude/skills/yolo

Other agents

`SKILL.md` is a single markdown file with structured instructions. Copy it into whatever prompt or skill format your agent uses, or paste it as a system prompt. It works with any coding agent that can read files, search the web, and write code.

How it works

  1. Understand (10 min): Scans your projects, notes, and recent work to learn your context
  2. Target (10 min): Researches opportunities (for build) or pain points (for refactor) or sources (for research)
  3. Commit (5 min): Lists 3 candidates, eliminates 2 with reasons, picks one
  4. Work (60-90 min): Does the actual work. Code, tests, iteration.
  5. Present: Summary of what was done, how to use it, what's next

Safety

All modes:

  • No outbound actions (no emails, messages, posts)
  • No paid services or purchases
  • No git push (you decide when to publish)
  • No global installs or syste