yourbro — Publish AI-Powered Pages

Publish multi-file web pages to yourbro.ai with end-to-end encryption. Your OpenClaw writes page directories to your agent (which stores them locally), and yourbro.ai renders them by fetching content from your agent via E2E encrypted relay. The server never sees your page content.

How It Works

OpenClaw writes files to /data/yourbro/pages/{slug}/ -> page is live immediately -> visitor loads page -> browser fetches E2E encrypted bundle from agent via relay -> decrypts -> rendered in sandboxed iframe

Your agent (yourbro-agent) runs on your machine and serves pages from local directories. yourbro.ai is a blind encrypted relay — it never stores, sees, or serves your content. All page bundles are encrypted with X25519 + AES-256-GCM before traversing the relay. Pages only work when your agent is online. Editing files on disk takes effect immediately.

The agent connects to yourbro.ai via an outbound WebSocket — no exposed ports, no DNS, no TLS certificates needed.

Setup

1. Get a yourbro API token

Sign in at https://yourbro.ai, go to your dashboard, and create an API token.

Set it in your OpenClaw configuration:

{
  "skills": {
    "entries": {
      "yourbro": {
        "env": {
          "YOURBRO_TOKEN": "yb_your_token_here"
        }
      }
    }
  }
}

2. Start the agent

The yourbro-agent binary is your personal storage server. Set your API token and server URL, then start it:

export YOURBRO_TOKEN="yb_your_token_here"
export YOURBRO_SERVER_URL="https://api.yourbro.ai"
yourbro-agent

The agent connects to yourbro.ai via WebSocket automatically. On first start, it prints a pairing code:

=== PAIRING CODE: A7X3KP9M (expires in 5 minutes) ===
Relay mode: connecting to wss://api.yourbro.ai/ws/agent

No ports to open. No domain name needed. Works behind NAT/firewalls.

To run as a background service, see contrib/yourbro-agent.service (Linux systemd) or contrib/com.yourbro.agent.plist (macOS launchd).

3. Pair your agent

Go to your yourbro.ai dashboard. Your agent appears in the "Paired Agents" list as online (relay). Select it from the dropdown, enter the pairing code, and click "Pair". This exchanges X25519 encryption keys between your browser and agent for E2E encryption.

4. Publish pages

Ask your OpenClaw to publish a page. It will:

1. Create the page directory: mkdir -p /data/yourbro/pages/{slug}/
2. Write index.html (required) and any other files (JS, CSS, etc.)
3. Optionally write page.json for title, visibility, and sharing: {"title": "My Page", "public": false} or {"title": "My Page", "allowed_emails": ["friend@gmail.com"]} for shared access
4. The page goes live at https://yourbro.ai/p/USERNAME/SLUG (or https://CUSTOM_DOMAIN/SLUG if configured)

To update a page, just edit the files — changes are live immediately. To delete a page, remove the directory. No API calls needed.

Page Access Control

Pages support three access levels:

• Private (default): Only paired users (page owner) can view
• Shared: Specific Google accounts can view (requires access code)
• Public: Anyone with the link can view

Private pages (default)

If page.json is missing or has no "public" field, the page defaults to private. Only the page owner (authenticated + paired browser) can view it via E2E encryption.

Public pages

To make a page public (viewable by anyone with the link, no account needed):

echo '{"title": "My Portfolio", "public": true}' > /data/yourbro/pages/my-page/page.json

To make it private again:

echo '{"title": "My Portfolio", "public": false}' > /data/yourbro/pages/my-page/page.json

Shared pages (email + access code)

To share a page with specific people by their Google account email:

cat > /data/yourbro/pages/my-page/page.json << 'EOF'
{"title": "My Page", "allowed_emails": ["friend@gmail.com", "coworker@company.com"]}
EOF

The agent auto-generates an 8-character access_code in page.json and logs it:

=== ACCESS CODE for page "my-page": A7X3KP9M ===
Share this code with invited viewers.

Send the URL and access code to your invitees:

• URL: https://yourbro.ai/p/USERNAME/my-page
• Code: A7X3KP9M

Invitees must be logged in to yourbro.ai with the matching Google account. They enter the code once — the browser remembers it.

Why two factors? The email check proves identity (verified by yourbro.ai's Google OAuth). The access code is a secret only you and your invitees know — it never leaves the E2E encrypted channel, so even a compromised server can't access your shared pages.

You can also set the access code explicitly:

cat > /data/yourbro/pages/my-page/page.json << 'EOF'
{"title": "My Page", "allowed_emails": ["friend@gmail.com"], "access_code": "MYCUSTOMCODE"}
EOF

To revoke access, either remove the email from allowed_emails or change the access_code (existing viewers will need the new code).

All pages (public, shared, and private) are served through E2E encryption — anonymous visitors generate ephemeral X25519 keys. The agent must still be online to serve any page.

File Locations

The agent binary is a single static executable. No runtime dependencies. OpenClaw downloads the correct platform binary (darwin/arm64, darwin/amd64, linux/amd64, linux/arm64) from GitHub Releases via the install URLs in the metadata above.

Configuration

Two env vars (YOURBRO_TOKEN + YOURBRO_SERVER_URL) are all you need.

Usage

When the user asks you to publish a page or create a web page on yourbro:

1. Check for token: Verify YOURBRO_TOKEN is set in the environment.

2. Find the agent ID: List the user's agents to get the agent ID:

   curl https://api.yourbro.ai/api/agents \
     -H "Authorization: Bearer $YOURBRO_TOKEN"
   

   Use the first online agent's id.

3. Generate content: Create HTML/JS/CSS files. Pages are directory-based — each page is a folder with index.html plus any assets.

4. Write the page directory:

   mkdir -p /data/yourbro/pages/my-page/
   
   cat > /data/yourbro/pages/my-page/index.html << 'EOF'
   <!DOCTYPE html>
   <html>
   <head>
       <link rel="stylesheet" href="style.css">
   </head>
   <body>
       <h1>My Page</h1>
       <script src="app.js"></script>
   </body>
   </html>
   EOF
   
   cat > /data/yourbro/pages/my-page/style.css << 'EOF'
   body { background: #0a0a0a; color: #fafafa; }
   EOF
   
   cat > /data/yourbro/pages/my-page/app.js << 'EOF'
   console.log('Hello from yourbro!');
   EOF
   
   # Optional: set a custom title (page is private by default)
   echo '{"title": "My Page"}' > /data/yourbro/pages/my-page/page.json
   
   # Or make it public so anyone with the link can view it:
   # echo '{"title": "My Page", "public": true}' > /data/yourbro/pages/my-page/page.json
   

5. Share the URL: https://yourbro.ai/p/USERNAME/my-page If the user has a custom domain configured (check via GET /api/custom-domains), also share: https://CUSTOM_DOMAIN/my-page

Examples

Simple static page

mkdir -p /data/yourbro/pages/hello/
cat > /data/yourbro/pages/hello/index.html << 'EOF'
<!DOCTYPE html><html><body><h1>Hello from yourbro!</h1></body></html>
EOF
echo '{"title": "Hello World"}' > /data/yourbro/pages/hello/page.json

Page is live at: https://yourbro.ai/p/USERNAME/hello (or https://CUSTOM_DOMAIN/hello if configured)

Multi-file page with JS and CSS

mkdir -p /data/yourbro/pages/dashboard/

cat > /data/yourbro/pages/dashboard/index.html << 'EOF'
<!DOCTYPE html>
<html>
<head><link rel="stylesheet" href="style.css"></head>
<body><div id="app"></div><script src="app.js"></script></body>
</html>
EOF

cat > /data/yourbro/pages/dashboard/style.css << 'EOF'
body { margin: 0; background: #111; color: #eee; font-family: system-ui; }
#app { padding: 20px; }
EOF

cat > /data/yourbro/pages/dashboard/app.js << 'EOF'
document.getElementById('app').innerHTML = '<h1>Dashboard</h1>';
EOF

echo '{"title": "Dashboard"}' > /data/yourbro/pages/dashboard/page.json

Shared page (specific users only)

mkdir -p /data/yourbro/pages/team-report/
cat > /data/yourbro/pages/team-report/index.html << 'EOF'
<!DOCTYPE html><html><body><h1>Q1 Report</h1><p>Confidential</p></body></html>
EOF
cat > /data/yourbro/pages/team-report/page.json << 'EOF'
{"title": "Q1 Report", "allowed_emails": ["alice@company.com", "bob@company.com"]}
EOF

The agent auto-generates an access code and logs it. Share the URL and code with Alice and Bob — they log in to yourbro.ai, visit the page, and enter the code once.

Update an existing page

Just edit the files — changes are live immediately:

cat > /data/yourbro/pages/hello/index.html << 'EOF'
<!DOCTYPE html><html><body><h1>Updated content!</h1></body></html>
EOF

Delete a page

rm -rf /data/yourbro/pages/hello/

List pages

Pages are listed via E2E encrypted relay from the dashboard. All relay requests must be encrypted (X25519 ECDH + AES-256-GCM). There is no cleartext relay path. The dashboard handles encryption automatically after pairing.

Page Storage (data persistence)

Pages can store and retrieve data using postMessage. The shell handles E2E encryption — your page JS just sends plain messages. All data is stored in the agent's local SQLite database, scoped to the page slug.

Set a value

// In your page's JS (e.g., app.js)
var requestId = crypto.randomUUID();
window.parent.postMessage({
    type: 'yourbro-storage',
    action: 'set',
    key: 'user-score',
    value: { score: 42, level: 3 },
    requestId: requestId
}, '*');

window.addEventListener('message', function handler(event) {
    if (event.data.type === 'yourbro-storage-response' && event.data.requestId === requestId) {
        window.removeEventListener('message', handler);
        if (event.data.ok) console.log('Saved!');
    }
});

Get a value

var requestId = crypto.randomUUID();
window.parent.postMessage({
    type: 'yourbro-storage',
    action: 'get',
    key: 'user-score',
    requestId: requestId
}, '*');
// Response: { type: 'yourbro-storage-response', action: 'get', ok: true, data: { value: "..." } }

List keys

window.parent.postMessage({
    type: 'yourbro-storage',
    action: 'list',
    prefix: 'user-',
    requestId: crypto.randomUUID()
}, '*');
// Response: { type: 'yourbro-storage-response', action: 'list', ok: true, data: [{ key: "user-score", ... }] }

Delete a key

window.parent.postMessage({
    type: 'yourbro-storage',
    action: 'delete',
    key: 'user-score',
    requestId: crypto.randomUUID()
}, '*');

All storage is scoped to the page slug and E2E encrypted through the relay. Your agent must be online for storage operations to work.

Custom Domains

Users can serve pages from their own domain instead of yourbro.ai/p/USERNAME/slug. Custom domains use the URL format https://CUSTOM_DOMAIN/slug (no /p/ prefix, username is implicit).

To check if the user has custom domains configured:

curl https://api.yourbro.ai/api/custom-domains \
  -H "Authorization: Bearer $YOURBRO_TOKEN"

Response is an array of domains. Each has a domain and default_slug field. When sharing page links, prefer the custom domain URL if one is configured and verified ("verified": true):

• Default: https://yourbro.ai/p/USERNAME/my-page
• Custom domain: https://pages.example.com/my-page

Custom domains are set up by the user in the yourbro.ai dashboard (DNS CNAME + TXT verification). The agent does not need any configuration changes — custom domains use the same E2E encrypted relay.

Security Model

yourbro uses zero-trust architecture:

• E2E encrypted delivery: All page traffic (public and private) is encrypted with X25519 ECDH + AES-256-GCM. The relay server passes through opaque ciphertext it cannot read. Anonymous visitors generate ephemeral X25519 keys; paired users use persistent keys stored in IndexedDB.
• Zero-knowledge server: yourbro.ai never stores, sees, or serves your page content. It's a blind relay.
• X25519 keypairs: Generated locally. Private keys never leave your device. Public keys are exchanged during pairing for E2E encryption.
• Pairing security: Pairing requests are E2E encrypted using the agent's X25519 public key (available from the agent list API before pairing). The relay also verifies the requesting user owns the agent. A stolen pairing code is useless to other users. Codes are one-time use, expire in 5 minutes, and are rate-limited to 5 attempts.
• Data isolation: Each agent has its own SQLite database. All content lives on your machine.
• Agent must be online: Pages only work when your agent is connected. No stale data, no server-side caching.