Operator Class

Free · Operator Kit

The operating documents behind real agent-built software.

This is not a prompt pack. It's the set of working documents an operator writes to direct AI agents through a real build — spec, operating manual, architecture request, review and QA checklists, launch and recovery playbooks.

Drop them in your repo. Fill the bracketed fields. They're the same kind of artifacts behind real agent-built systems — the difference between software that ships and a demo that breaks on the second click.

Get instant access

Enter your email and the full Kit opens on the next page — nothing to wait for.

Free, instant access. See our privacy notice.

What's inside

13 operating documents

Start with the three-doc spine, then use the rest as each stage arrives. Every document comes with a filled worked example from a real, shipped build.

First 30 minutes

  1. 01

    AGENTS.md

    Set the agent's character before it helps you think.

  2. 02

    PRODUCT_SPEC.md

    Interview the rough idea into a real PRD and first slice.

  3. 03

    DEFINITION_OF_DONE.md

    Define what done means before the agent starts building.

Start here

Map the Kit, then set the agent's operating contract.

  • Operator Kit — Start Here

    What the Kit is, the Operator Loop it maps to, and the order to run the documents in.

  • Agent Operating Manual (AGENTS.md)

    The standing rules every agent reads every session — tools, verification gates, and the facts it keeps forgetting.

Spec

Decide what should exist, and how you'll know it works.

  • Product / Spec Template

    A PRD without the bloat — including a paste-ready Spec Interview that turns a rough idea into a buildable first slice.

  • Definition of Done & Acceptance

    The reusable bar every slice must clear — so 'done' means the same thing every time.

Orchestrate

Slice the work and hand the agent one piece at a time.

  • Scope & Milestone Plan

    Slice a build into vertical, shippable steps so the agent can't run off a cliff between checkpoints.

  • Task Brief

    The atomic handoff: one slice, scoped, with acceptance criteria, verify commands, and a stop condition.

  • Architecture Request

    Ask the agent for architecture options and trade-offs instead of accepting whatever it types first.

  • UX / UI Direction Brief

    Direction so the interface reflects your taste, not the agent's default SaaS template.

Review

Judge the output without pretending to be a senior engineer.

  • Code Review Checklist (for non-engineers)

    What you can and should check on every diff — without pretending to be a senior engineer.

  • Agent Recovery & Debugging Playbook

    What to do when the agent drifts, breaks the build, or confidently ships nonsense.

Ship

Prove it works, then deploy without chaos.

  • Browser QA Checklist

    Prove the build actually works where users live — the part demos always skip.

  • Launch & Deployment Checklist

    Ship to production without chaos — env, domains, email, and a rollback mindset.

Grow

Measure the funnel and decide what to build next.

  • Analytics & Attribution Starter

    Instrument the funnel before launch so you learn from real traffic instead of guessing.