Build with AI
These docs are written to be read by AI coding agents as well as people. The dStorage SDK is new and its API changes between versions, so an agent working from what it remembers will often get it wrong. Point it at these docs instead, and it works from the current API.
Point your agent at the docs
Every page is also published as plain Markdown for agents:
llms.txt— an index of every page, with links to their Markdown versions. Start here.llms-full.txt— the whole docs in one file, for agents that work best with everything in context at once.- A Markdown copy of each page — the page URL with
.mdinstead of.html, e.g./docs/guide/mock-adapters.md.
Some agents' fetch tools can't read the .md copies. If yours can't, give it the normal page URL or llms-full.txt instead.
Add the docs to your project's agent instructions
Most coding agents read a project instructions file on every run: AGENTS.md (Codex, Cursor and others), CLAUDE.md (Claude Code) or .cursor/rules. Add this to yours, so the agent checks these docs whenever it touches dStorage code, without you having to remind it:
## dStorage SDK
This project uses @dstorage-tech/dstorage-sdk. Its API changes between versions, so don't rely
on prior knowledge of it. Before writing or changing dStorage code, read the docs:
- Docs index: https://dstorage.pro/docs/llms.txt
- Full docs in one file: https://dstorage.pro/docs/llms-full.txt
Check the SDK version installed in package.json, and flag anything in the docs that doesn't
match it.Ask about any page
The Copy page for AI button in the top bar copies the current page as Markdown. Its menu can also open the page in Claude or ChatGPT, ready for you to ask questions about it — handy for the FAQ and Features pages.
A prompt for every guide
Each guide starts with a ready-made prompt that has your agent follow it in your own project, ending with a check that it worked. They're collected here in the same order as the guides, each building on the one before:
Prefer a working app to start from? Clone dstorage-starter-template and run the prompts against it.
Tips
- Review what the agent changes, especially anything touching keys, passwords or tokens.
- Never paste a
ds_*secret token into a prompt or browser code. Keep tokens in environment variables; in the browser, use a scoped JWT token (see Managed Payments). - Steps that need you — approving wallet prompts, passkey ceremonies, creating a dStorage Pro token — can't be done by the agent. The prompts tell it to stop and ask you at those points.