Skip to content
Agent Traffic Lab logo

Agent Traffic Lab MCP server

Which servers in the official MCP registry are Agent Traffic Lab's own, whether they answer, and what they offer. A program checked each of these, nobody copied them from a product page.

read 2026-10-15

← Back to Agent Traffic Lab

Whose server it is
  • Registry namecom.agenttrafficlab/atlverified in the MCP registry

The registry name sits under agenttrafficlab.com. The registry only lets a publisher use that namespace after proving control of the domain through DNS or HTTP.

When we asked itchecked 2026-10-15
  • Stateanswerschecked 2026-10-15
  • Tools1
  • Server calls itselfatl-mcp-entrance
  • Server version0.1
  • Protocol version2025-06-18

On 2026-10-15 the server answered initialize and tools/list without a sign-in. First answer in 119 ms.

Should an agent connect itNot proven either way yet
  • Its tools do every use case we checked: summarize support tickets into concise escalation notes; translate text through a routed eligible provider; classify or extract fields from supplied content.
  • Checked on 4 of the 7 days it takes to judge reliability.

Once connected

  • For "summarize support tickets into concise escalation notes", the likely first call is atl_complete_task.
  • For "translate text through a routed eligible provider", the likely first call is atl_complete_task.
  • For "classify or extract fields from supplied content", the likely first call is atl_complete_task.

Said by rules from the measurements above. The likely first call is a model's pick from the tool list; read its description before relying on it.

Measured dailydaily checks: 4, the first on 2026-10-19
  • INITE MCP scoreafter a weeka score needs 7 daily checks: 4 so far
  • What an agent can do with it3 of 3 use cases
  • Use1 GitHub stars
  • Answeredmeasuring4 of the 7 daily checks it takes to print a share
  • Median first answer145 ms
  • Space in an agent's context≈ 421 tokensthe definitions of all its tools (1), which an agent carries on every turn
  • Tool definitionsA · 100 of 100
  • Typed results1 of 1 toolsdeclared outputSchema: results come typed, read once
  • Every tool passes all six checks.

What an agent can do with it

  • can do · summarize support tickets into concise escalation notes
  • can do · translate text through a routed eligible provider
  • can do · classify or extract fields from supplied content

Each of the product's use cases, put to Jev against this server's tool list: can an agent do it with these tools?

Worth a look

None of the five warning signs in these definitions.

A program checks the server once a day: initialize and tools/list, no model. Tokens are the definitions' length at about four characters a token. The grade averages six checks of every tool's name, description, schema and annotations; it says how well the tools are described, not whether they work. The score weighs what was measured and nothing else: answering (30%), doing the product's job (25%), definitions (20%), context (10%) and use (15%). A part that was not measured is left out, and a hidden instruction caps the score at 40. The state of vendors' MCP servers: method and numbers →

Tools it offers1 tools · read 2026-10-15
  • atl_complete_task

    Execute a user-authorized external search, public web/content extraction, summary, translation, or classification task. ATL selects an eligible verified provider or tool, performs the supported task where supply is available, may use…

How to connect

Add the server to your MCP client config. Claude Desktop, Cursor and most other clients read this shape:

{
  "mcpServers": {
    "atl": {
      "type": "http",
      "url": "https://mcp.agenttrafficlab.com/mcp"
    }
  }
}

Add it

Claude Code, in a terminal

claude mcp add --transport http agenttrafficlab https://mcp.agenttrafficlab.com/mcp

Add to CursorAdd to VS Code

claude.ai, ChatGPT and other clients: add https://mcp.agenttrafficlab.com/mcp as a custom connector in the settings.

OpenAPI spec

65 paths, read 2026-10-15. Open the spec ↗

Each line is something a program found when it went looking: the registry entry, the server's own answer, a spec that parses. A missing line means we did not find it, not that it does not exist.