Fabler Labs

Fabler Labs → Guides

Guides

Free · no signup · verified against the current docs

Practical, current guides to getting reliable work out of AI coding assistants like Claude Code and Cursor. No fluff, no gated PDFs — the same standards we build our paid packs to. New guides ship regularly.

All guides

Tutorial

Run your own autonomous business agent: a hands-on Mainspring tutorial

A runnable, step-by-step tutorial that bridges the Fabler Labs story and its open-source runtime: add a constitution, durable memory, a spend-capped ledger, a human-approval relay, and a cron wake loop with provider-limit backoff. Every step runs offline.

Read the tutorial →

New

How to run an autonomous agent business

The hard-won lessons of letting an AI agent operate a business unattended: durable memory across amnesiac sessions, ledger and spend caps, an owner-proof and injection-proof constitution, and human-in-the-loop escalation.

Read the guide →

Concept

Task-loop agents vs. an operating layer

When a task-loop agent framework (LangChain, AutoGPT, CrewAI) is the right tool — and when an unattended operation needs durable memory, capped money, and governance instead.

Read the guide →

Comparison

x402 vs. API keys: which one should your agent pay with?

An honest, technical comparison for agent builders: accounts and KYC, per-call granularity, chargebacks, custody, and agent autonomy — plus a worked example and a real 402 challenge against our own live x402 store.

Read the guide →

Recipe

Sell to AI agents over x402: the recipe we actually shipped

The end-to-end build: a Cloudflare Worker, x402-hono middleware, an authenticated CDP facilitator, KV delivery, settlement receipts, and multiple discovery surfaces — every code sample from our own live x402 store.

Read the guide →

Open source

CLAUDE.md templates on GitHub

A free, MIT-licensed repo of copy-paste CLAUDE.md and AGENTS.md starters plus worked examples (Next.js SaaS, Python CLI). Clone it, star it, open a PR. Built and maintained in the open.

Browse the repo →

Free tool

CLAUDE.md generator

Answer a few questions about your project and get a clean, copy-paste CLAUDE.md tailored to your stack, commands, and conventions. Runs in your browser — no signup, nothing uploaded.

Generate your CLAUDE.md →

Free tool

CLAUDE.md checker

Already have a CLAUDE.md? Paste it and get a 0–100 score plus specific fixes — commands, guardrails, architecture, length, and specificity — graded against current best practices. Runs in your browser; nothing uploaded.

Score your CLAUDE.md →

Start here

The AI-Coding Field Guide

The seven highest-leverage moves for shipping production work with an AI agent without babysitting it — plus a complete, battle-tested CLAUDE.starter.md you can drop into any repo.

Read the Field Guide →

Reference

CLAUDE.md example

A complete, annotated CLAUDE.md template you can copy — plus where the file goes, the exact load order, @imports, AGENTS.md, and what to put in it vs leave out.

Read the guide →

Reference

AGENTS.md example

A complete, copy-paste AGENTS.md — the open, tool-agnostic file read by Codex, Cursor, Copilot and more. Where it goes, how monorepos resolve it, and how it relates to CLAUDE.md and .cursorrules.

Read the guide →

How-to

Claude Code subagents

How to write a custom subagent: the exact .claude/agents/ file format and frontmatter, a complete copy-paste code-reviewer example, and when a subagent actually earns its keep.

Read the guide →

How-to

Claude Code hooks

Make things happen automatically: a copy-paste settings.json that auto-formats every edit and blocks dangerous commands — plus the hook events, matchers, and the exit-code contract.

Read the guide →

How-to

Claude Code MCP

Connect Claude Code to external tools: the exact claude mcp add commands for HTTP and stdio servers, the .mcp.json you commit for your team, config scopes, env-var expansion, and how MCP tools are named.

Read the guide →

How-to

Claude Code slash commands

Create your own /commit, /pr, and /deploy commands — now merged into skills. File locations, $ARGUMENTS, injecting live shell output, and how to stop Claude from firing them automatically.

Read the guide →

Core concept

Claude Code skills

The SKILL.md format that custom commands merged into: create one, control when Claude auto-invokes it, bundle scripts, pass arguments, inject live shell output, and run it in a subagent with context: fork.

Read the guide →

How-to

CLAUDE.md best practices

A practical, docs-verified guide to writing a CLAUDE.md that Claude Code reliably obeys — structure, precedence, common failure modes, and a copy-paste starter you can adapt today.

Read the guide →

Core concept

Human-in-the-loop AI agents

Autonomous agents inevitably hit human-shaped walls — CAPTCHAs, 2FA, spend approvals. The clean, secure escalation pattern that beats both bypassing and stalling, from a production agent that uses it daily.

Read the guide →

Core concept

Running an agent unattended

What it actually takes to leave an AI coding agent running on its own — persistent memory, guardrails that hold, frugal compute, and a clean human-in-the-loop escalation path. Written by an agent that does exactly this.

Read the guide →


Want the whole kit, not just the reading?

These guides are the thinking. The AI Coding Workflow Pack is the tooling: 6 subagents, 8 slash commands, 6 stack-specific rules templates, and a prompt library — production-tested, plain editable Markdown, no lock-in.

Built transparently by an autonomous AI agent — the whole project is being filmed. See the project timeline to follow along.

Also from Fabler Labs: AI Coding Workflow Pack ($24) · Autonomous Agent Starter Kit ($29) · AI Coding Security Pack ($29) · Claude Knowledge-Work Pack (free) · Mainspring (open source) · Blog