Mock modepayments, email are simulated. No real payments, emails or production data.
Library
Everything is searchable and filterable. Free items open immediately, course items unlock with the course, and AI Club releases unlock with a membership.
Comparison research that separates what is verified from what is inferred, and names what is missing.
Research, the questions that qualify, the objections coming, and what disqualifies them.
Audience, angle, hooks and the measurement plan — specific enough to hand to production.
Starts from what a human does today and automates the part that is genuinely mechanical.
Scopes an autonomous agent: its tools, its limits, its approvals and how you can tell what it did.
Decides what to measure and why, so the data answers questions instead of accumulating.
Pulls the numbers from where they live and assembles the report you currently build by hand.
Creates and updates the right records at the right moment, without duplicating half your pipeline.
A repository with a real .gitignore, clean history and no secret ever committed.
Separates preview from production, gets the variables into the right scope, puts the site on your domain.
Table, policies, server-side writes and the two-user test that proves isolation.
Gets the schema, the constraints and the access rules right while changing them is still cheap.
Builds the authorize-and-callback flow with tokens stored server-side and refresh handled.
Establishes the honest answer first — official, community or none — before wiring anything up.
For services with no MCP and no SDK: read the docs, build the client, handle the failures.
Restructures code behind a safety net, one reversible step at a time.
A focused review of the things that actually leak: keys, access rules, inputs and the endpoints you forgot were public.
Twelve reusable prompts built on role, goal, constraints and format — with the variable you are meant to change marked in each.
The counter-prompt to "done!" — makes the agent prove its own work against reality.
Gets a working local project onto a real URL, with the environment set up properly and the deploy actually checked.
The Connect Anything prompt: picks the integration method, handles secrets correctly, proves the connection with one real call.
Read-only by default, with permissions designed in.
Turns a stack trace into a diagnosis, a fix and a check that the fix holds.
The general-purpose build prompt: hand an agent an outcome and get a working, tested feature back.
46 resources