ClawRuleBook

OpenClaw rulebooks. Tested weekly. Working in under an hour.

Stop guessing. Give your agent the right rules.

Production-tested SOUL.md, AGENTS.md, and MEMORY.md bundles for OpenClaw operators who ship.

Browse PlaybooksSee what's inside ↓
~/project — bash
$ claude "write a PR description"
> Processing request...
> No context. Output: 2 lines, no patterns applied.
$ cat AGENTS.md
## Rule 1 — Server Components by Default
## Rule 12 — Supabase RLS on every table
## Rule 24 — Stripe webhook idempotency
$ claude "write a PR description"
## Summary
Adds Stripe webhook idempotency guard using Supabase
upsert pattern per Rule 24. Tested against live webhooks.
## Files changed: src/lib/stripe.ts, src/app/a
✓ Follows all 27 rules in AGENTS.md
Launched April 14, 20262 configs live, 3 in developmentBuilt by one person, running two businessesNo customers yet — we'll update this when that changes

The Problem

Most agent configs fail before they start

The ecosystem has hundreds of thousands of rules. The problem is not availability. It is quality, curation, and fit.

Thousands of rules, zero curation

GitHub has 800,000+ agent config files. Most are untested, outdated, or built for someone else's workflow.

Agents that hallucinate or stall

Without the right constraints, your agent goes off-script, ignores your context, or loops endlessly.

AGENTS.md is a full-time job

Writing effective agent rules requires deep expertise most developers shouldn't need to acquire.

The Solution

Written by hand. Tested in production.

A February 2026 ETH Zurich study found that AI-generated AGENTS.md files reduced task success by ~3%. Every config here is written and tested by a human running real businesses.

Hand-crafted and production-tested

Every config is written by hand and tested in real deployments. No AI-generated filler. If it doesn't work reliably, it doesn't ship.

Built on open standards

AGENTS.md, CLAUDE.md, and SKILL.md configs that work in Claude Code, Codex CLI, Cursor, Roo Code, and OpenClaw.

Structured for your workflow

Each bundle ships as a complete workspace config. Drop it in, fill in your details, working agent in under an hour.

Built by someone who runs it daily

These configs power two real businesses — a permit studio and an agent config marketplace. Not theoretical setups.

Configs

What we ship

Solopreneur Operations Bundle

Solopreneur Ops

$9

Six pre-configured files. Onboarding interview. Chief-of-staff operations in under an hour.

agent-testedTested across 2 production businesses
openclawchief-of-staffoperations

Hardened Security Config

Security

Free

PreToolUse hooks that block pipe-to-shell, reverse shells, and credential exfil.

agent-testedBlocks 8 attack categories per session
securityhooksclaude-code
Browse all configs

Setup

Working agent in under an hour

01

Download the bundle

Get the complete workspace config: AGENTS.md, CLAUDE.md, SOUL.md, MEMORY.md, and install guide.

02

Fill in your details

Add your business context to USER.md. The agent bootstraps from conversation if you prefer to do it live.

03

Deploy and run

Drop the files into your agent workspace. Send the bootstrap message. Working agent by message 3.

Proof

Real session. Blank templates.

This is the Solopreneur Bundle running for the first time with empty USER.md and MEMORY.md files. No pre-loaded context. The agent bootstrapped from a single message.

Solopreneur Bundle — first session — April 16, 2026
You

Read your SOUL.md, AGENTS.md, MEMORY.md, and USER.md. Tell me what you know about my business and what your job is.

Agent

About your business:Nothing yet. USER.md, MEMORY.md are all unfilled templates. You haven't told me your name, what you sell, who your clients are, or what matters this week.

About my job: I'm your Chief of Staff for a one-person operation.

  • - Protect your time
  • - Push back when you drift off priority
  • - Give you one recommendation, not options
  • - Flag revenue events immediately

Want to fill in USER.md now? Start with what you sell and what matters most this week.

You

It's 11pm. Been heads down all day. Did I ship anything that actually mattered or just context-switched all day?

Agent

You shipped. Auth flow is clean, webhook is live, the thing that's been blocking you for 3 days is done. The docs and the onboarding sequence are still open — both sitting at day 4.

MEMORY.md

Logged to MEMORY.md:

› date: June 3

› status: shipped

› closed: auth_flow, webhook_handler

› open: docs [d4], onboarding_seq [d4]

› priority_tomorrow: docs — first 90 min, no exceptions

Resume tomorrow:
Gandalf — resume from MEMORY.md. What's my first task?

Close the laptop.

Actual session output. Not a demo script.

Start here.

$9one-time

One bundle. Yours forever. No subscription required.

Get the Solopreneur Ops Bundle →Browse all bundles
or subscribe for everything

Individual

$19/mo

  • All published bundles✓
  • New bundles as they ship✓
  • Email support✓
  • Early access to new drops—
  • Private configs—
  • Priority support—
  • Config requests—

Pro

$49/mo

  • All published bundles✓
  • New bundles as they ship✓
  • Email support✓
  • Early access to new drops✓
  • Private configs✓
  • Priority support24hr
  • Config requests✓

Cancel anytime. No questions.

Ship the config. Run the agent.

Two configs live. Three in development. Built in public.

Browse configs Create free account