AI AGENTS

Give your AI agents eyes on the web

Most agent frameworks can call an HTTP endpoint as a tool. Snapshot Site gives that tool call real browser rendering — screenshot, visual diff, and AI analysis — instead of raw scraped HTML the agent has to interpret blind.

MCP-ready
Hosted server, OAuth, zero setup
Plain HTTP
Works with any agent framework
3 tools
screenshot, analyze, compare
Install:curl -X POST https://api.prod.ss.snapshot-site.com/api/v1/screenshot
Auth:x-snapshotsiteapi-key: YOUR_API_KEY
Get started for free
Agent tool call
Good fits
Agents that verify a deploy or a form submission visually
Research agents that need a real rendering of a page, not raw HTML
QA and monitoring agents comparing before/after states
RAG pipelines that need visual context alongside text

Two ways to plug in

If your agent runtime already speaks MCP (Claude Code, Cursor, Claude Desktop, and others), the hosted MCP server is the fastest path. If it doesn't, or you're building a custom agent loop, calling the plain HTTP API as a tool works the same way any other function call would.

1

MCP for MCP-native clients

Discover, authorize, and call screenshot/analyze/compare as tools with no custom integration code.

2

Plain HTTP for everything else

Any agent framework that can make an HTTP request can call the same endpoints directly as a tool.

3

Real rendering, not scraped HTML

The agent gets what a browser actually rendered — JavaScript, layout, and all — instead of raw markup it has to reconstruct.

4

Visual diff as a verification step

Agents that make a change to a page can immediately verify it visually with a compare call, not just check an HTTP status.

Quick start

Wire the tool into your agent

1

MCP-native agent: connect to the hosted MCP server, no key management needed

2

Custom agent loop: register a tool that POSTs to /api/v1/screenshot, /v3/analyze, or /v3/compare

3

Return the response image URL and metrics to the agent's context

4

For verification steps, prefer /v3/compare over a single screenshot

Example tool call

Compare

Agent verifies a page change

An agent that just made an edit calls compare to confirm the visual result before reporting success.

curl --request POST \
  --url https://api.prod.ss.snapshot-site.com/api/v3/compare \
  --header 'Content-Type: application/json' \
  --header 'x-snapshotsiteapi-key: YOUR_API_KEY' \
  --data '{
    "before": {"imageUrl": "https://cdn.example.com/before.png"},
    "after": {"url": "https://staging.example.com/page", "width": 1440, "fullSize": true},
    "threshold": 0.1
  }'

Related resources

Snapshot Site MCP ServerUsing Snapshot Site MCP with Claude Desktop and CursorSnapshot Site is Now Available Inside ClaudeAPI Documentation