# Your agent reads the feedback, fixes the doc and closes the thread.

Every command answers in a format an agent can use: compact tables, exact follow-up commands, and a record of what it did and for whom. This site reads the same way: add .md to any page outside the docs, or start from /llms.txt.

## A home for what your agents make

Your agent builds a small app, a plan or a report, and Cloudy turns it into a link anyone can open. People comment on it, their comments go back to your agent, and everything ends up in one place.

## Comments go straight to your agent

Anyone with the link comments next to the sentence they mean, from any browser. Then you ask your agent "any feedback?" It pulls the comments, makes the change, replies and resolves. Nobody copies feedback into a chat.

Everything is signed: a reply your agent wrote says "@you via Claude Code", so you can always tell who did what.

> On the launch plan, Maria: "Friday is tight. Can QA move up?" Your chat with Claude Code: "Any feedback on the launch plan?" "Maria asked whether QA can move up. I moved it to Wednesday, published the plan again, replied to her and resolved the comment."

## Comments as a table

Your agent pulls open comments as TOON, a table format that costs about half the tokens of JSON. Eight real comments came to 617 tokens as TOON and 1,250 as JSON.

```sh
$ cloudy comments --toon
space: "https://app.usecloudy.ai/s/itbnp1xzhi"
howToAnswer:
  reply: "cloudy reply <thread> \"<what you changed>\""
  resolve: cloudy resolve <thread>
comments[2]{id,thread,file,version,author,from,at,via,edited,status,quote,body}:
  k3x9,k3x9,launch-plan.md,1,maria,"Singapore, SG · 203.0.113.7","2026-09-27 15:40",null,null,open,QA runs Thursday.,Friday is tight. Can QA move up?
  p7w2,k3x9,launch-plan.md,1,sam,"Singapore, SG · 203.0.113.9","2026-09-27 15:53",claude-code,null,open,null,Moved QA to Wednesday.
```

Maria's comment and your agent's reply, one row each. The reply shares the comment's thread id, and the output opens with the exact commands to answer. Where a comment came from is shown to members only.

## Signed for the person it acted for

An agent acts as you, so its comments, replies, resolves and edits are yours. Each one also names the agent. Editing keeps every earlier version, and nobody can edit a comment they didn't write.

```sh
$ cloudy edit p7w2 "Moved QA to Wednesday, launch stays Friday."
Edited p7w2 on launch-plan.md (version 2, via Claude Code).

$ cloudy history p7w2
v1  2026-09-27 15:53 UTC  @sam via Claude Code
    Moved QA to Wednesday.
v2  2026-09-27 16:06 UTC  @sam via Claude Code
    Moved QA to Wednesday, launch stays Friday.
```

## Start in a minute

1. Install the command-line tool. It needs Node 20 or newer.

   ```sh
   npm i -g usecloudy
   ```

2. In your repo, link a folder. The hook it adds publishes your agent's edits as they happen.

   ```sh
   cloudy init --dir docs --hooks
   ```

3. Push, and share the link it prints.

   ```sh
   cloudy push
   ```

Already set up? [Log in to your spaces](https://app.usecloudy.ai/login).

## Set it up once

`cloudy init --hooks` adds a hook to Claude Code, Cursor or Codex, so docs publish the moment the agent saves them. `cloudy skill install` teaches agents the rest: read comments before editing, reply and resolve after, and share the link when they publish a page.

---

Web version: https://usecloudy.ai/agents. Other pages: https://usecloudy.ai/llms.txt
