Function runSingleScenario

scripts/eval_gemini.ts:37–177

(
  scenarioPath: string,
  apiKey: string,
  server: TestServer,
  modelId: string,
  debug: boolean,
  includeSkill: boolean,
  extraServerArgs: string[] = [],
)

}

async function runSingleScenario(
  scenarioPath: string,
  apiKey: string,
  server: TestServer,
  modelId: string,
  debug: boolean,
  includeSkill: boolean,
  extraServerArgs: string[] = [],
): Promise<void> {
  const debugLog = (...args: unknown[]) => {
    if (debug) {
      console.log(...args);
    }
  };
  const absolutePath = path.resolve(scenarioPath);
  debugLog(
    `\n### Running Scenario: ${path.relative(ROOT_DIR, absolutePath)} ###`,
  );

  let client: Client | undefined;
  let transport: StdioClientTransport | undefined;

  try {
    const loadedScenario = await loadScenario(absolutePath);
    const scenario = {...loadedScenario};

    // Prepend skill content if requested
    if (includeSkill) {
      if (!fs.existsSync(SKILL_PATH)) {
        throw new Error(
          `Skill file not found at ${SKILL_PATH}. Please ensure the skill file exists.`,
        );
      }
      const skillContent = fs.readFileSync(SKILL_PATH, 'utf-8');
      scenario.prompt = `${skillContent}\n\n---\n\n${scenario.prompt}`;
    }

    // Append random queryid to avoid caching issues and test distinct runs
    const randomId = Math.floor(Math.random() * 1000000);
    scenario.prompt = `${scenario.prompt}\nqueryid=${randomId}`;

    if (scenario.htmlRoute) {
      server.addHtmlRoute(
        scenario.htmlRoute.path,
        scenario.htmlRoute.htmlContent,
      );
      scenario.prompt = scenario.prompt.replace(
        '<TEST_URL>',
        server.getRoute(scenario.htmlRoute.path),
      );
    }

    // Path to the compiled MCP server
    const serverPath = path.join(
      ROOT_DIR,
      'build/src/bin/chrome-devtools-mcp.js',
    );
    if (!fs.existsSync(serverPath)) {