Every build, written up
Every guide comes from something I actually built and use. Copy the workflow, install the skill, get the result.
Fact-check your content with an agent before you post
Before a post ships, an agent fact-checks every claim against primary sources and flags overclaims. One wrong claim costs trust you cannot buy back.
Give every repo a CLAUDE.md runbook any AI can follow
A repo only you can operate is a repo only you can maintain. A root CLAUDE.md runbook fixes that: golden rules, jobs as literal steps, gates before deploy.
Give your agents lanes: listen, speak, answer
Agent chaos ends when every tool gets one lane: listening tools never publish, one tool speaks, and anything touching private DMs stays read-only and manual.
Capture ideas from your phone into your second brain
Ideas die between your phone and your desk. My loop: one chat message, an agent drops it in the vault inbox, a nightly agent compiles your second brain.
Give your AI a memory that survives restarts
Chat memory dies with the session. Files survive. One markdown file per fact plus an index your agent loads at start gives any AI a permanent memory.
Keep Claude focused: hand subagents the heavy work
Long Claude sessions drown in file reads and logs. Subagents do the heavy work in their own context window and return only the conclusion, so your chat stays sharp.
Let agents keep your goals file current
Goal docs rot because updating them is manual. Make yours derived files: you capture raw thoughts, a scheduled agent recompiles goals and priorities.
Let the top model plan and a cheaper model build
Usage limits and surprise bills come from running your strongest model on everything. The fix: the top model plans and reviews, a cheaper model builds.
Make AI prove its work before you call it done
AI says done long before the work is done. Define the acceptance check in the prompt and make the agent produce the proof: tests, screenshots, a green build.
Make Claude Code follow your rules with hooks
CLAUDE.md and skills only suggest behavior. Hooks enforce it at fixed lifecycle points. Copy a hook that blocks .env edits even when permissions are bypassed.
Put your AI on a schedule and let it work while you sleep
The jump from chatbot to AI system is a schedule. Copy my pattern: a job that wakes an agent with a written goal, plus spend caps, least privilege, and logs.
Run a team of AI agents from one prompt
One prompt in Claude Code can fan out into a team of AI agents. The strongest model plans and reviews, cheap models do the grunt work. Copy the setup.
Set spend caps before your AI agents run overnight
An always-on agent with a paid API key is a surprise bill waiting to happen. Two layers of spend caps bound the damage before your first overnight run.
Stop tool shopping: pick one tool per layer
Productivity tool debates are really capture UI debates. Split your stack into three layers, pick one tool per layer, and let agents move work between them.
Triage your Instagram DMs with AI that never replies
Let an AI agent read and sort your Instagram DMs, flag who needs a reply, and never send a word. The read-only triage pattern that keeps your replies human.
Turn any workflow you repeat into a Claude Code skill
If you explain a workflow to Claude twice, make it a skill. What a SKILL.md is, why the description line matters, and a prompt that writes your first one.
One dashboard for every agent you run
Background agents pile up across your machines and you lose track. A local board reads each machine's session state from git and shows what needs you.
Build real AI agents without writing code
You do not write agent code. You describe the job in plain English, one orchestrator prompt fans out to worker agents, and a reviewer checks the result.
Make Fable 5 rebuild your Claude Code setup before it goes
Fable 5 access is temporary. I pointed it at 3,339 of my own transcripts, and it rebuilt my Claude Code setup so every future session is cheaper. Here is the prompt.
Save your Claude context before it runs out
When a chat gets long, Claude starts forgetting. One small habit hands you everything you need to continue in a fresh chat with zero lost progress.
Give your AI agents their own task queue
Your agents sit idle between your prompts. Mine pull work from their own board: a planner writes tickets, a dispatcher runs up to 3 every wake.
Build a RAG coach for your community
How I shipped a chatbot that answers only from my community's own courses, with magic-link login and daily spend caps. Live in production.
Render Instagram carousels from one command
One typed idea becomes branded 1080x1350 slides. Claude writes the copy, a renderer draws the pixels, I pick the hook. That split keeps it on brand.
Put your competitor research on autopilot
A scheduled Claude Code agent reads your competitors' reels and comments, then reports which hooks work, what fans keep asking, and what to adapt.
Give every idea a nightly AI co-founder
Ideas die in note apps. Mine get a scheduled agent that researches and scores them night by night, then writes a PRD for the winner. Copy the setup.
Ask your notes anything with a local RAG
One script, SQLite, and a free local embedding model. Ask your whole notes vault questions from the terminal, with ranked citations. No API keys.
Build a second brain that maintains itself
You drop messy notes into one inbox folder. A scheduled agent files them into a clean linked wiki while you sleep. One rule keeps it from breaking.
Nothing matches that filter yet. New guides land every week.
Want to build this yourself?
Join Claude Code Academy: courses, build alongs, and builders turning AI into real systems. And I post a new build on Instagram nearly every day.