How to use AgentViva

One tool at a time: the dashboard

For anyone. No code.

1. Open Monitors and press Add a monitor.
2. Choose what to check: a website, a JSON health address, or an AI answer (Pro).
3. Paste the address (example.com is enough) and press Add monitor.
4. The first check runs within one check period (a day for a website, unless you choose faster). Press Check now on the monitor's page to check at once.

Your whole list: a catalog

For directories, marketplaces and anyone with many tools.

1. Publish one file on your site with every tool: an id, a name and an address. JSON or CSV.
2. Open Catalog in the menu, paste the link to the file, and press Save and check the list.
3. That is all. Every day we read the list again: a new tool gets a monitor, a renamed or moved tool is updated, and a tool you removed is paused, with its history kept.

EXAMPLE FILE (JSON)

[
  { "id": "notion", "name": "Notion", "url": "https://www.notion.so" },
  { "id": "geondex", "name": "Geondex", "url": "https://geondex.com" }
]

We also read slug for the id, title for the name, and website or link for the address. A CSV needs a header row.

From your code: the API

For developers.

1. Open API in the menu and press Make a new API key. Copy it: it is shown only once.
2. Send it in the Authorization header of every request, as below.
3. Everything else is in the API reference.

ADD A MONITOR

curl -X POST https://agentviva.dev/v1/monitors -H "Authorization: Bearer av_live_..." -H "content-type: application/json" -d '{"type":"page","name":"My tool","target":{"url":"https://example.com"}}'

CONNECT A CATALOG

curl -X PUT https://agentviva.dev/v1/catalog -H "Authorization: Bearer av_live_..." -H "content-type: application/json" -d '{"url":"https://example.com/tools.json"}'

READ WHAT CHANGED

curl "https://agentviva.dev/v1/changes?since=0" -H "Authorization: Bearer av_live_..."

LIST WHAT DOES NOT WORK

curl "https://agentviva.dev/v1/monitors?state=down,degraded&sort=problems" -H "Authorization: Bearer av_live_..."

About your API key

Your agent uses the same key as your code.

• Make it once. Open API in the menu and press Make a new API key.
• Shown once. We show the full key only when you make it, and keep only a hash of it. Save it in a password manager.
• One key for the team. Your code, your scripts and your AI agents all use this same key.
• Lost it? Make a new one. The old key stops at once, so put the new key everywhere the old one was.
• Keep it secret. Anyone with the key can add, change and delete your monitors.

Claude Code: the plugin (easiest)

It brings the MCP connection and the skill (our rules for the agent).

1. Install it from your terminal:

   claude plugin marketplace add AgentViva/docs
   claude plugin install agentviva@agentviva

2. Add your key. In Claude Code, type /plugin, open the Installed tab, choose AgentViva, then Configure options, and paste the key.
3. Restart Claude Code, then type /mcp. agentviva should show 9 tools. Only 1 tool means the key is not set yet. "Failed" means the key is wrong.
4. Ask in plain words, as above.

Installed it before? Update it first: claude plugin marketplace update agentviva, then claude plugin update agentviva@agentviva.

No key yet? Skip step 2. The plugin still has one tool, lookup_tool: it answers "is Notion up?" from the tools AgentViva already watches.

Without the plugin

Claude Code, Cursor, Codex or any other MCP client.

CLAUDE CODE, ONE LINE

claude mcp add --transport http agentviva https://agentviva.dev/mcp --header "Authorization: Bearer av_live_..."

ANY OTHER MCP CLIENT: A REMOTE (HTTP) SERVER

Address: https://agentviva.dev/mcp
Header:  Authorization: Bearer av_live_...

claude.ai and Claude Desktop connectors do not work yet: they need a sign-in method (OAuth) we have not built.