Recipe library

Automations you can actually read

Every recipe here is one file. You can read it before you run it, change it with a text editor, and run it again tomorrow without an AI being involved at all. Copy any of them.

The library

Pick the result you want. Each page shows the whole file.

News digest

Wake up to a Hacker News digest on your desktop

Every morning there's a markdown file on your desktop with today's top stories. You read it in two minutes.

Read the recipe →
Research

Get the trending AI repos delivered, without opening GitHub

A markdown report on your desktop listing the AI repositories gaining stars right now.

Read the recipe →
Briefing

One briefing instead of five open tabs

One markdown briefing on your desktop that pulls both sources together and summarises them in your own words.

Read the recipe →
Developer

Stop writing up what you did today

A clean markdown changelog of today's work, written for humans, saved to your desktop.

Read the recipe →
Start here

The simplest recipe there is: ask a local model, save the answer

A model on your own machine answers a prompt, and the answer lands in a file on your desktop.

Read the recipe →
Market watch

Know when a competitor changes their pricing

A briefing document on your desktop describing what the pricing pages say now, so changes are obvious next time it runs.

Read the recipe →
Tracking

Build your own price history, one row a day

A CSV on your desktop that gains one row every time the recipe runs. Open it in Excel and chart it.

Read the recipe →
Reporting

A market report that writes itself every morning

A formatted market report on your desktop, plus the raw data saved alongside it so you can audit any number in the report.

Read the recipe →
Diagnostics

Find out what's listening before you debug for an hour

A diagnostic brief on your desktop telling you what responded and what didn't.

Read the recipe →
Files

Find out what's actually in that folder

A report listing what's in the workspace, saved where you can read it.

Read the recipe →
Files

Start every project the same way, in one command

A new project workspace with your structure and a filled-in manifest, created on your desktop.

Read the recipe →
Offline

Summarise a document without it ever leaving your machine

A written summary on your desktop, produced without the document touching the internet.

Read the recipe →
Any model

The same recipe, whichever model you point it at

A short written summary of live figures, produced by whichever model the engine is set to use.

Read the recipe →
Research

The same tracker, with a model explaining why each repo matters

A brief on your desktop where each trending repo comes with a sentence on what it actually does.

Read the recipe →
Alerts

Get pinged on Discord when Bitcoin crosses your line

A message in your own Discord channel, on your phone and your computer, only when the price goes above or below the numbers you set.

Read the recipe →
Reliability

Stop one failed API call from killing the whole job

Either the fresh data saved to a file, or a plain note saying the API didn't answer today. Never a silent crash.

Read the recipe →
Files

See what changed in a folder today

A short note listing every file in that folder that was created or changed today.

Read the recipe →
Chaining

Run your whole routine with one command

One recipe that runs the others in sequence and tells you when they're all done.

Read the recipe →
Web

Turn a web page into clean text you can use

A markdown file holding the readable text of the page, with the markup stripped out.

Read the recipe →
Human checkpoint

Let the automation stop and ask you when it hits a login

A browser window opens on the page that needs you. You sign in or solve the challenge, confirm, and the recipe carries on from there.

Read the recipe →

Why a file instead of a chat

Ask a chat assistant to do a job and it does it once, charging you for every token, and tomorrow you ask again. A recipe is written once and then it's a program: it runs on a schedule, it does the same thing every time, and if it has no model step, no model is involved when it runs. You can read exactly what it touches before it runs.

Using an AI assistant? Ask your AI to check out adeptuscamini.com.