About & methodology

AgentRecipes is a library of practical "how to automate X with AI agents" recipes. Each recipe describes a real automation that someone built and wrote about in public, using tools such as n8n, Make, Zapier Agents, Lindy, Claude and ChatGPT. It currently has 39 recipes in 8 categories, last updated 2026-10-15.

Where recipes come from

We find use cases mainly through Agent Use Cases (agentusecases.ai), a hand-reviewed directory of real AI agent use cases, and its RSS feed. Every recipe page credits the Agent Use Cases entry it was found through.
Every recipe links to the original write-up: a forum post, blog article, repository, workflow template or news article by the person who built it. If we can't find a working original source, we don't publish the recipe.
Publication dates are shown as listed by Agent Use Cases and may only be accurate to the month. Many sources have no date, so we leave it blank.
Source sites in this edition: Artificial Corner (Substack), DEV Community, DX blog, GitHub, Hacker News, Labellerr blog, Medium, PCWorld, Reddit, Simone Smerilli's blog, Substack, Tom's Guide, X, XRay.Tech blog, n8n Community forum, n8n workflow library.

How recipes are written

The problem statement, steps, tool list and caveats are summarised in our own words from the source. We don't copy text wholesale. For the full details, screenshots and any code, read the original.
"What the builder reported" repeats only what the source says, attributed to its author. We have not tested or verified those claims. Where a source gives no numbers, we say so instead of filling the gap.
We don't publish invented statistics, user counts or fake workflow files. If a source shares a workflow export, the source link is where to get it.
Every Prompt template is an original template written by AgentRecipes to help you get started. It is not the source author's prompt. Fill in the placeholders and test it on your own data before relying on it.
Tool names, prices and plan limits change. Check the vendor's current pricing and terms before you build.

Safety notes

Most recipes keep a human approval step before anything is sent, paid or published. We recommend you keep it too, at least until you trust the automation.
Give agents the least access they need (read-only first), and be careful with email, bank and customer data.

Updates

New recipes are added from the Agent Use Cases feed and other public sources, and source links are re-checked regularly. Spotted a broken link, a mistake or a recipe that misrepresents its source? Please tell us and we'll fix or remove it.

AgentRecipes has one sponsor slot for automation and AI tools. Sponsorship details are coming soon. Sponsored placements will always be labelled and won't change which recipes we publish or how we describe them.

Data access

The full dataset is at /api/recipes.json (our summaries and templates under CC BY 4.0; source write-ups belong to their authors). LLM-friendly summary at /llms.txt. Browse by category or platform.