MCP tool

probierz_evaluate_talk

SIDE-EFFECTING: judge a rendered talk's slide pictures and speaker script against the documents it presents through the authenticated model router, criterion by criterion (source fidelity, argument, slide design, human authorship, delivery), and write an immutable JSON verdict that passes only when every criterion is met.

Input schema#

JSON Schema
{
  "type": "object",
  "properties": {
    "slidesPath": {
      "type": "string",
      "description": "Directory of slide-<number>.png pictures written by lem-cli talk render --slides."
    },
    "scriptPath": {
      "type": "string",
      "description": "Speaker script Markdown written by lem-cli talk render --notes."
    },
    "sourcePaths": {
      "type": "array",
      "items": {
        "type": "string"
      },
      "description": "Documents the talk must stay true to: Markdown, text, TeX or PDF; at least one."
    },
    "rubricPath": {
      "type": "string",
      "description": "Optional JSON rubric; uses the panel-talk rubric by default."
    },
    "model": {
      "type": "string",
      "description": "Vision-capable model ID; defaults to PROBIERZ_TALK_MODEL."
    },
    "agentId": {
      "type": "string",
      "description": "Agent identity for subscription routes; the secret comes from PROBIERZ_MODEL_AGENT_SECRET."
    },
    "outputPath": {
      "type": "string",
      "description": "Optional destination ending in .json; existing evidence is never overwritten."
    }
  },
  "required": [
    "slidesPath",
    "scriptPath",
    "sourcePaths"
  ]
}

Source snapshot#

Generated from `probierz mcp` at https://github.com/wisent-ai/probierz.git revision `f65b755da397f2ec7435eb80ed722aa6017437c1`.

The server reports version `0.1.5`; binary SHA-256: `d5398832d3ab52c541b1efe979730dea62941ac6197898a20fccfc5d3fca44df`; tools/list SHA-256: `e0ac65d9045c1fdeb35ed9b386f1532ec72ee0dca4586e624c817fec41bbaf86`.