Build with AI

Connect AI agents like Claude Code, Cursor, and Codex to 3,100+ apps through the MindCloud Universal API. One skill file teaches your agent to discover apps, read action schemas, and run actions.

Your AI agent can use the Universal API on its own. It needs two things: a MindCloud API key and instructions.

Skills

We publish the instructions as an agent skill, one markdown file that teaches the whole loop: find an app, list its actions, read a schema, resolve IDs by name, run the action. Follow these steps to use it in your favorite agent:

Create your MindCloud account, then create a key in API Keys and export it where your agent runs:
```
export MINDCLOUD_API_KEY=your_key_here
```
A read_only key lets the agent browse apps and action schemas. A run_workflows key also lets it resolve lookups and run actions.
Install the MindCloud skill:
```
npx skills add https://mindcloud.co/skills/SKILL.md
```
You can read the skill first if you want to see what your agent will follow.
Ask your agent for something real:
```
Ask me which apps I use at work, then show me what you can do with them through MindCloud.
```
The agent matches your tools against the 3,100+ supported apps, checks which ones you already connected, and pulls real data from those. For the rest it sends you to the connections page. From there, ask for anything: pull records, cross-reference two apps, or send a Slack message.

More for machines

MindCloud API skill covers the account side: workflows, runs, connections, members, and usage.
OpenAPI spec describes every v2 endpoint and needs no key to fetch.
llms.txt orients a model on MindCloud as a whole.
Browse apps has a reference page for every app, with arguments and examples per action.