Skip to content
WeaversDocs
Browse documentation

Weavers / MCP documentation

Understand an experiment

Read recorded holdout facts for an experiment you can view.

Read onlyget_experiment_facts

Required permission: weavers.experiments.read

Try asking

Explain the recorded holdout facts for this Weavers experiment: <run ID>.

How it works

  • Use a UUID run ID from Weavers. This external tool accepts UUIDs even though some other Weavers interfaces accept broader run identifiers.
  • Access follows the experiment visibility rules in Weavers. A missing or inaccessible run returns not_found.
  • The facts array can vary by run. Null values mean evidence is unavailable; truncated indicates bounded trade processing.

Inputs

FieldRequiredType & constraints
runIdRequiredUUID string
View complete input JSON schema
Input schema

{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "type": "object",
  "properties": {
    "runId": {
      "type": "string",
      "format": "uuid",
      "pattern": "^([0-9a-fA-F]{8}-[0-9a-fA-F]{4}-[1-8][0-9a-fA-F]{3}-[89abAB][0-9a-fA-F]{3}-[0-9a-fA-F]{12}|00000000-0000-0000-0000-000000000000|ffffffff-ffff-ffff-ffff-ffffffffffff)$"
    }
  },
  "required": [
    "runId"
  ],
  "additionalProperties": false
}

What comes back

FieldDescription
runIdString. Requested experiment identifier.
scopeLiteral "holdout".
truncatedBoolean. Whether the evidence was bounded.
facts[]Objects with key (string), label (string), and value (number, string or null).

This table describes the decoded tool result. MCP wraps it in text content and, for object results, structuredContent.

Worked example

Fictional data. Replace identifiers, revisions and hashes with values from your own Weavers workflow.

Request

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "get_experiment_facts",
    "arguments": {
      "runId": "33333333-3333-4333-8333-333333333333"
    }
  }
}
Decoded result

{
  "runId": "33333333-3333-4333-8333-333333333333",
  "scope": "holdout",
  "truncated": false,
  "facts": [
    {
      "key": "status",
      "label": "Status",
      "value": "completed"
    }
  ]
}

If the request fails

  • Sign in and grant weavers.experiments.read before calling this tool.
  • Use resources you can access and refresh revision information when it changes.
  • Check result.isError even when HTTP returns 200.
Find the error and next step →