← WebMCP Directory

Tool tree

webmcp-research-auditor.vercel.app

Health & EducationTools verified
Website
Evidence-first comparison of machine-learning experiment runs.
Explore capabilities
    get_current_comparisonAnswer

    Read the two runs currently selected by the researcher and the decision question visible on the page. This never changes records.

    View tool JSON
    {
      "name": "get_current_comparison",
      "kind": "answer",
      "impl": "imperative",
      "description": "Read the two runs currently selected by the researcher and the decision question visible on the page. This never changes records.",
      "inputSchema": {
        "type": "object",
        "properties": {
          "include": {
            "type": "array",
            "items": {
              "enum": [
                "metrics",
                "configs",
                "artifact_summaries"
              ]
            }
          }
        },
        "additionalProperties": false
      },
      "page": "/demo"
    }
    run_comparability_auditAction

    Run deterministic checks against the current selected pair. The result is a temporary page preview; it does not create a finding or investigation.

    View tool JSON
    {
      "name": "run_comparability_audit",
      "kind": "act",
      "impl": "imperative",
      "description": "Run deterministic checks against the current selected pair. The result is a temporary page preview; it does not create a finding or investigation.",
      "inputSchema": {
        "type": "object",
        "properties": {
          "selectionDigest": {
            "type": "string",
            "description": "Digest returned by get_current_comparison."
          },
          "question": {
            "type": "string",
            "description": "Optional decision question."
          }
        },
        "required": [
          "selectionDigest"
        ],
        "additionalProperties": false
      },
      "page": "/demo"
    }
    show_finding_evidenceAction

    Open the page's side-by-side source evidence for a temporary or saved finding. This changes only the visible panel.

    View tool JSON
    {
      "name": "show_finding_evidence",
      "kind": "act",
      "impl": "imperative",
      "description": "Open the page's side-by-side source evidence for a temporary or saved finding. This changes only the visible panel.",
      "inputSchema": {
        "type": "object",
        "properties": {
          "findingId": {
            "type": "string"
          },
          "evidenceRefIds": {
            "type": "array",
            "items": {
              "type": "string"
            }
          }
        },
        "required": [
          "findingId"
        ],
        "additionalProperties": false
      },
      "page": "/demo"
    }
    stage_challenge_revisionAction

    Stage a reversible interpretation preview using researcher-provided context. A human must confirm before anything enters the investigation history.

    View tool JSON
    {
      "name": "stage_challenge_revision",
      "kind": "act",
      "impl": "imperative",
      "description": "Stage a reversible interpretation preview using researcher-provided context. A human must confirm before anything enters the investigation history.",
      "inputSchema": {
        "type": "object",
        "properties": {
          "findingId": {
            "type": "string"
          },
          "researcherContext": {
            "type": "string",
            "minLength": 1
          }
        },
        "required": [
          "findingId",
          "researcherContext"
        ],
        "additionalProperties": false
      },
      "page": "/demo"
    }
    stage_resolution_planAction

    Stage an editable reevaluation plan. This does not save or approve the plan and grants no permission to execute experiments.

    View tool JSON
    {
      "name": "stage_resolution_plan",
      "kind": "act",
      "impl": "imperative",
      "description": "Stage an editable reevaluation plan. This does not save or approve the plan and grants no permission to execute experiments.",
      "inputSchema": {
        "type": "object",
        "properties": {
          "investigationId": {
            "type": "string"
          },
          "constraints": {
            "type": "object"
          }
        },
        "required": [
          "investigationId"
        ],
        "additionalProperties": false
      },
      "page": "/demo"
    }
    review_investigation_readinessAction

    Review the live evidence, interpretation, and plan state, then rank the highest-value improvements. This is read-only and cannot save, confirm, approve, or execute anything.

    View tool JSON
    {
      "name": "review_investigation_readiness",
      "kind": "act",
      "impl": "imperative",
      "description": "Review the live evidence, interpretation, and plan state, then rank the highest-value improvements. This is read-only and cannot save, confirm, approve, or execute anything.",
      "inputSchema": {
        "type": "object",
        "properties": {},
        "additionalProperties": false
      },
      "page": "/demo"
    }
    get_decision_briefAnswer

    Return the bounded canonical decision brief and rendered Markdown. This does not initiate a download or change any record.

    View tool JSON
    {
      "name": "get_decision_brief",
      "kind": "answer",
      "impl": "imperative",
      "description": "Return the bounded canonical decision brief and rendered Markdown. This does not initiate a download or change any record.",
      "inputSchema": {
        "type": "object",
        "properties": {},
        "additionalProperties": false
      },
      "page": "/demo"
    }