Guides library · 14 working guides

Practical AI guides you can actually check

A working library for people who use AI seriously: task briefs, handoffs, redaction, human review, local-first tools, and the LYGO reference material behind them.

No course, no paid tier, and no promise that one prompt fixes everything. Each guide is written to be checked against a tool or a public record, and each one states its limits.

By Justin Helmer · reviewed 18 October 2026

Advertisement

Start here

Operator craft

Reference

What this library is

This is a small publisher's working library, not a content farm. Each page exists because a real task kept going wrong: a handoff lost a file path, a redaction missed a key, a “done” claim could not be re-checked, a free tool hid its limits. The guide is the repair note, written so the next person does not repeat the failure.

The library is deliberately narrow. It covers the operator craft around AI work: how to brief an agent, how to move work between sessions, how to keep secrets out of a prompt, how to review output before publishing, and how to tell a public reference from a canonical record. It also covers the LYGO material that the tools are built around.

How to choose a guide

Start with the job in front of you, not with the size of the library.

How these guides are made

A guide starts with a task or a failure, not a keyword. I test the steps against the tool or the public record, then write the page so another person can repeat them. AI may help with drafting or editing; the byline is human, and the human is responsible for the published claim.

Checkable claims get a link, a command, or a date. Claims that cannot be checked are written as opinion. When a page changes materially, the visible review date changes too. The full process is on the editorial standards page.

What makes a guide worth reading

A guide earns its place when it answers a question the tool cannot answer by itself: when to trust a result, how to check a claim, what to redact, how to hand work to another person. The useful pages here include a worked example, a failure mode, and a way to check the result. A page that only repeats the marketing copy is a draft.

Corrections and review

If a guide is wrong, the correction is part of the library. Send the page URL and the sentence through the contact page. The page changes, and the review date changes with it. A quiet deletion is not a correction.

FAQ

Are the guides free?

Yes. They are free to read. Optional ads and donations support hosting; they do not gate the pages.

Do I need an account?

No. The guides, the public ledgers, and the free tools do not require an account. A specific third-party tool may require its own key; the guide says so.

Where should I start?

Start with the question you already have. If you only want the map, read the lattice overview. If you want the reference, read the protocol guide. If you want the craft, start with task briefs.

Can I reuse a guide?

Quote a short passage with attribution and a link. Do not republish the full text or sell it. Ask first for anything beyond a quotation.

Publisher portrait
Justin Helmer

Publisher of chatagent.ca · known as Excavationpro / Lightfather in LYGO projects. Writes practical guides on AI personas, agent handoffs, and free local-first tools. About · Contact

More from the workshop: Excavationpro · Asian Coastline · BPMfinder · Justin Helmer · DeepSeek Oracle · Eternal Haven · All tools