Autonomous agents

Agents, REST API & MCP

One key for everything: agent apps, the REST API v1, and the native MCP server.

Create an agent app, get a key, and call every CVAI tool over plain HTTP or through the Model Context Protocol server — same keys, spend caps, and audit logging across both.

Get an API key All surfaces

Setup ~5 min

Plain REST + JSON plus a native MCP endpoint — call from any language or MCP client.

Per-app cvai_ak_ keys with isolated spend limits and rate limits.

Sync or async runs, with webhooks to deliver results back to your agent.

Best for: Teams building their own apps and builders wiring CVAI into autonomous agents or AI coding assistants.

Get started

Follow these steps to get up and running.

Create an agent app

Create an app from your signed-in developer dashboard — give it a name and description.

# Create your agent app in the dashboard:
# https://communityvision.ai/dashboard/developer

Generate an API key

From your app in the dashboard, generate a key. The raw key is shown only once — store it somewhere safe.

# Generate a key for your app in the dashboard:
# https://communityvision.ai/dashboard/developer
#
# The raw key (cvai_ak_...) is shown only once — copy it now.

Call a tool over HTTP

Authenticate every request with your key as a Bearer header. Sync or async mode.

curl -X POST https://communityvision.ai/api/v1/tools/spec_review/runs \
  -H "Authorization: Bearer cvai_ak_your_key" \
  -H "Content-Type: application/json" \
  -d '{
    "input": {
      "sourceText": "# My PRD\n\nThis product will...",
      "analysisDepth": "full"
    },
    "mode": "sync"
  }'

# Poll an async run:
# GET https://communityvision.ai/api/v1/tool-runs/{runId}

Or wire up the MCP server

In Claude Desktop, Cursor, or any MCP client, add the CVAI server with the same key.

{
  "mcpServers": {
    "cvai": {
      "url": "https://communityvision.ai/api/mcp",
      "headers": {
        "Authorization": "Bearer cvai_ak_your_key"
      }
    }
  }
}

Set limits & webhooks

In the developer dashboard, open your agent app to set spend caps, the rate limit, and a webhook URL (these settings are managed from the signed-in dashboard, not with the API key).

# Manage daily/monthly spend caps, rate limit,
# and your webhook URL from your agent app:
# https://communityvision.ai/dashboard/developer

Helpful resources

Developer dashboard Full API documentation Automation & embeds

Other surfaces

CVAI meets you everywhere you work. Explore the other integrations.

Command Line (CLI)

Audit specs and plans in a single command, right from your terminal.

VS Code extension

Audit the spec open in your editor and see findings as inline squiggles.

GitHub Action

Gate pull requests on an audit of the specs and plans they touch.

Slack app

Mention @CVAI in any channel to audit a document, right inside Slack.

Browser (extension & WebMCP)

Audit any page with one click, or let AI browsers call CVAI tools right on the page.

Mobile handoff

A bundle of design tokens, assets, and specs to build the mobile app with zero design drift.

Automation & embeds

Get results via signed webhooks, and embed your reports into any page.

Agents, REST API & MCP — frequently asked questions

How do I call CVAI tools from my own agent or app?

Create an agent app in the developer dashboard, generate a per-app cvai_ak_ key, and POST to the REST API v1 (for example /api/v1/tools/spec_review/runs) with the key as a Bearer header. The same key also works with the native MCP server at /api/mcp from Claude Desktop, Cursor, or any MCP client.

Do CVAI agent API keys have spend and rate limits?

Yes. Each agent app gets its own cvai_ak_ key with isolated daily and monthly spend caps, a rate limit, and an optional webhook URL — all managed from the signed-in developer dashboard, with the same spend caps and audit logging applied across REST and MCP.