webmcp-lab.vercel.app

{
  "name": "list_targets",
  "kind": "answer",
  "impl": "imperative",
  "description": "List the sites this Lab can evaluate and the task fixtures it can run against them. Call this first: run_eval needs an id from each list.",
  "inputSchema": {
    "type": "object",
    "properties": {}
  }
}

{
  "name": "run_eval",
  "kind": "act",
  "impl": "imperative",
  "description": "Run one task fixture against one target site through that site's WebMCP tools, then check the site's own state to decide whether the task really happened. Returns a verdict and a run id.",
  "inputSchema": {
    "type": "object",
    "properties": {
      "targetId": {
        "type": "string",
        "enum": [
          "field-notes",
          "field-notes-v2",
          "field-notes-fixed"
        ],
        "description": "Which site to evaluate"
      },
      "fixtureId": {
        "type": "string",
        "enum": [
          "save-neurodivergent-posts"
        ],
        "description": "Which task to run"
      }
    },
    "required": [
      "targetId",
      "fixtureId"
    ]
  }
}

{
  "name": "inspect_run",
  "kind": "answer",
  "impl": "imperative",
  "description": "Explain one run in full: the tools the target published, any contract problem found before execution, every tool call, and every assertion with its expected and actual value. Use this after a failing run_eval to find out what to fix.",
  "inputSchema": {
    "type": "object",
    "properties": {
      "runId": {
        "type": "string",
        "description": "A run id from run_eval or inspect_run. Omit it to use the most recent run."
      }
    }
  }
}

{
  "name": "rerun_eval",
  "kind": "act",
  "impl": "imperative",
  "description": "Run the same target and task again and compare the result to the earlier run. Use this after changing the site to prove the fix held, or to catch a regression.",
  "inputSchema": {
    "type": "object",
    "properties": {
      "runId": {
        "type": "string",
        "description": "A run id from run_eval or inspect_run. Omit it to use the most recent run."
      },
      "targetId": {
        "type": "string",
        "enum": [
          "field-notes",
          "field-notes-v2",
          "field-notes-fixed"
        ],
        "description": "The build to run now. Defaults to the same one as the earlier run; name a different build to check whether a fix held."
      }
    }
  }
}

{
  "name": "export_report",
  "kind": "act",
  "impl": "imperative",
  "description": "Write a Markdown report for one run and download it, so the result can be pasted into a pull request or an issue. Returns the same Markdown.",
  "inputSchema": {
    "type": "object",
    "properties": {
      "runId": {
        "type": "string",
        "description": "A run id from run_eval or inspect_run. Omit it to use the most recent run."
      }
    }
  }
}