INNER CODE UNIT · TypeScript

verifyEvalRunTool

HoangNguyen0403/agent-skills-standard · mcp/src/tools/index.ts:841

export async function verifyEvalRunTool(
  args: { run_id?: string },
  ctx: ToolContext,
): Promise<ToolResult> {
  const runIds = args.run_id ? [args.run_id] : listEvalRuns(ctx.projectRoot);

  if (runIds.length === 0) {
    return {
      content: [
        {
          type: "text",
          text: "No eval runs found under `benchmarks/evals/runs/`. Run the `evals-run` workflow first (see docs/EVALS.md) to produce one.",
        },
      ],
    };
  }

  const outcomes = runIds.map((id: string) =>

View source record →