Package dev.dokimos.mcp
Class ToolHandlers
java.lang.Object
dev.dokimos.mcp.ToolHandlers
Implements the four MCP tool handlers for the dokimos evaluation framework.
-
Constructor Summary
ConstructorsConstructorDescriptionToolHandlers(ResultStore store, com.fasterxml.jackson.databind.ObjectMapper json) -
Method Summary
Modifier and TypeMethodDescriptionio.modelcontextprotocol.spec.McpSchema.CallToolResulthandleCompareRuns(Map<String, Object> arguments) Compares two runs viaRunComparison: per-evaluator deltas, pass-rate test, per-item diffs.io.modelcontextprotocol.spec.McpSchema.CallToolResulthandleGetFailingQueries(Map<String, Object> arguments) Returns failing queries from a run, filtered by score threshold.io.modelcontextprotocol.spec.McpSchema.CallToolResulthandleListExperiments(Map<String, Object> arguments) Lists past evaluation runs with optional filtering.io.modelcontextprotocol.spec.McpSchema.CallToolResulthandleRunEvaluation(Map<String, Object> arguments) Runs an evaluation: loads a dataset, calls an LLM, evaluates results, and persists the run.
-
Constructor Details
-
ToolHandlers
-
-
Method Details
-
handleRunEvaluation
public io.modelcontextprotocol.spec.McpSchema.CallToolResult handleRunEvaluation(Map<String, Object> arguments) Runs an evaluation: loads a dataset, calls an LLM, evaluates results, and persists the run. -
handleListExperiments
public io.modelcontextprotocol.spec.McpSchema.CallToolResult handleListExperiments(Map<String, Object> arguments) Lists past evaluation runs with optional filtering. -
handleCompareRuns
public io.modelcontextprotocol.spec.McpSchema.CallToolResult handleCompareRuns(Map<String, Object> arguments) Compares two runs viaRunComparison: per-evaluator deltas, pass-rate test, per-item diffs. -
handleGetFailingQueries
public io.modelcontextprotocol.spec.McpSchema.CallToolResult handleGetFailingQueries(Map<String, Object> arguments) Returns failing queries from a run, filtered by score threshold.
-