@turbot/powerpipe-mcp/dist/utils/powerpipe_run.js

Powerpipe MCP server to run benchmarks, detections and controls using AI.

import { formatCommandError } from "./command.js";
import { logger } from "../services/logger.js";
/**
 * Handle the output from powerpipe run commands that return JSON.
 *
 * Powerpipe run commands (benchmark, control, detection) may exit with non-zero status
 * even when they successfully execute and return valid JSON output. This happens when
 * the checks detect issues:
 *
 * Exit codes:
 * - 0: Success - ran successfully with no alarms or errors
 * - 1: Alarm - completed successfully but there were one or more alarms
 * - 2: Error - completed but one or more control errors occurred
 *
 * In these cases, we want to return the JSON output (which contains the failure details)
 * rather than treating it as a command error.
 *
 * @param error The error from executeCommand
 * @param cmd The command that was executed
 * @returns Formatted response with either parsed JSON output or error message
 */
export function handlePowerpipeRunOutput(error, cmd) {
    // If we have stdout, try to return it as valid output even if command failed
    if (error instanceof Error && 'stdout' in error) {
        const cmdError = error;
        if (cmdError.stdout) {
            try {
                // Validate it's JSON
                const parsed = JSON.parse(cmdError.stdout);
                return {
                    content: [{
                            type: "text",
                            text: JSON.stringify({
                                output: parsed,
                                debug: {
                                    command: cmd
                                }
                            })
                        }]
                };
            }
            catch (parseError) {
                logger.error('Failed to parse powerpipe output:', parseError);
            }
        }
    }
    return formatCommandError(error, cmd);
}
//# sourceMappingURL=powerpipe_run.js.map