The connected library for everything you feed your AI — here’s the whole toolkit, from capturing your prompts, rules, and skills to plugging them live into your editor, your agents, and your whole company.
Prompts, rules, docs, templates, configs, skills — typed, structured, and easy to find.
Prompts, agent/Cursor rules, markdown docs, mcp.json configs, skills, and prescriptive templates — each with the right format and file extension — plus a File type for everything else (ignore files, scripts) that keeps its real name and extension. Define your own item types too.
Turn prompts into reusable templates with [fields]. Click a field, drop in a value, and copy the finished result — so you reuse the prompt, not the rewrite.
Drag in your files or a .zip, or pull a whole GitHub repo — public, or private with a token we use for that one import and never store. Your prompts, rules, skills, docs, templates and configs land auto-detected as the right type, ready to review — no retyping.
Categories, tags, favorites and pins, plus search by title, content, author, or who last edited it. Open libraries as tabs you can drag to reorder and pin the ones you live in.
Deactivate an item you're not using — a work-in-progress skill, or the few shared items you don't need. It dims out of your way and your AI tools skip it. It's per-user: only you, nobody else is affected.
Your library doesn't just sit there — it plugs into your editor and your agents.
Connect your library once with the AIStorageDepot MCP server and your agent can reach in on demand inside Claude Desktop, Cursor, Cline, Windsurf, VS Code, Claude Code, and Codex. Just ask in plain language — no copy-paste. Try it with no account first — run it token-free for a free, read-only sample of the prebuilt library.
Over MCP, your prompts show up as native slash-commands across Claude Code, Cursor, VS Code, Windsurf, and Codex — they take your input as arguments and the AI fills their [fields]. Want your skills as typed commands too? One optional pull installs them as local files, no token needed.
One click opens any item in Claude, ChatGPT, or Perplexity — prefilled and ready to run. Add a wrapper like “Enhance this prompt:” or turn a one-off into a reusable [bracketed] template.
Download any item as a clean Markdown, JSON, or YAML file — then drop it into your repo as the file your tool reads: AGENTS.md, .cursor/rules/*.mdc, mcp.json.
A clean API and personal access tokens back everything — reach your library from any browser (laptop or phone), your editor, your agents, or a script. One source of truth.
Edit freely — every change is tracked, and nothing gets clobbered.
Every edit is tracked with complete history, and you can restore any previous version in one click. An overzealous edit is never a disaster.
Optimistic locking catches two people editing the same item at once and warns instead of clobbering — and you can see who else is in an item while you're there.
Anything you delete waits in the Trash for 30 days, restorable in a click — so a stray delete is a quick undo, not a loss.
Copy an item from a shared library and it remembers where it came from: when the original moves ahead your copy flags it, one click brings it current, and a Lineage tab draws the whole family tree — original, every copy you can see, and which ones are behind.
Give everyone the same playbook, with the controls a company needs.
A company-wide library every employee sees, plus per-team libraries — so everyone pulls from the same approved prompts, rules, and skills. Roles, invites, and central admin included.
Copy or move any personal item into a team or company library in a click — keep your original, or hand it off. Push company items down to teams, promote team items up to the company library, and seed a new team straight from the prebuilt examples — one item at a time or a whole selection in bulk. Define item types once and use them across every team.
Admins can lock items so they stay clean: a soft lock a manager can lift, or a hard lock only an admin over that person can clear. Lock and unlock in bulk at every level.
A company owner can export every library across the whole organization as a single .zip, organized into folders exactly as it appears in the app.
Your prompts and docs are your working knowledge. We treat them that way.
Encryption in transit and at rest, optional two-factor authentication, bcrypt-hashed passwords, and automatic lockout after failed sign-ins — turn on 2FA and a stolen password isn't enough.
A self-hosted edition deploys on your own Azure or Docker with your own PostgreSQL — prompts never leave your network. One-time purchase, yours forever, no phone-home; and a company owner's export imports straight into a self-hosted install, so starting in the cloud never locks you in.
VPN/IP allowlisting to fence your org onto your network, an audit log of every administrative change — roles, invites, membership, access settings — instant member revocation, and an email alert on every new-device sign-in.
Billing runs entirely on Stripe. Your card number goes to Stripe's vault, never our servers or our database.
Automatic encrypted backups as a disaster-recovery net, a 90-day grace window if you go over a limit or a plan lapses, and we never purge accounts for inactivity.
Give any item a public read-only link — strangers see a clean view with no account needed, and teammates who sign in open the real item. Links stay active until you revoke them, and you can revoke one anytime.
Download any item as a clean file, pull your whole library over the API and MCP, or permanently delete your account yourself from Settings — your data leaves whenever you want.