Parallels RAS MCP Server

Instructions

Implementation Reference

• src/tools/help-support.ts:31-38 (handler)

  The ras_support_info tool handler that makes an async GET request to '/api/help-and-support/support' endpoint and returns the support information as formatted JSON, with error handling using sanitiseError.

  async () => {
    try {
      const data = await rasClient.get("/api/help-and-support/support");
      return { content: [{ type: "text" as const, text: JSON.stringify(data, null, 2) }] };
    } catch (err) {
      return { content: [{ type: "text" as const, text: sanitiseError(err, "Failed to retrieve support info") }], isError: true };
    }
  }

• src/tools/help-support.ts:19-39 (registration)

  The registration of ras_support_info tool with the MCP server, including title, description, annotations (read-only), empty input schema (no parameters), and the handler function.

  server.registerTool(
    "ras_support_info",
    {
      title: "Support Information",
      description:
        "Get Parallels RAS support information, including support contact details, " +
        "product version, build number, and support entitlement. Use this to check " +
        "the installed RAS version, verify support status, or gather information " +
        "for a support ticket.",
      annotations: READ_ONLY_ANNOTATIONS,
      inputSchema: {},
    },
    async () => {
      try {
        const data = await rasClient.get("/api/help-and-support/support");
        return { content: [{ type: "text" as const, text: JSON.stringify(data, null, 2) }] };
      } catch (err) {
        return { content: [{ type: "text" as const, text: sanitiseError(err, "Failed to retrieve support info") }], isError: true };
      }
    }
  );

• src/tools/help-support.ts:29-29 (schema)

  The input schema definition for ras_support_info tool - an empty object indicating the tool takes no input parameters.

  inputSchema: {},

• src/client.ts:43-54 (helper)

  The sanitiseError helper function used by ras_support_info handler to sanitize error messages by removing auth tokens, passwords, and truncating long responses to avoid leaking internal details.

  function sanitiseError(err: unknown, context: string): string {
    const raw = err instanceof Error ? err.message : String(err);
    // Remove anything that looks like a token or password value
    let sanitised = raw
      .replace(/auth_token[=:]\s*\S+/gi, "auth_token=[REDACTED]")
      .replace(/password[=:]\s*\S+/gi, "password=[REDACTED]");
    // Truncate excessively long API response bodies
    if (sanitised.length > 500) {
      sanitised = sanitised.substring(0, 500) + "... (truncated)";
    }
    return `${context}: ${sanitised}`;
  }

• src/client.ts:128-166 (helper)

  The rasClient.get() method used by ras_support_info to make authenticated GET requests to the RAS API, handling lazy authentication, automatic retry on 401 errors, and request timeouts.

  async get(path: string): Promise<unknown> {
    // Ensure we have a valid session
    if (!this.authToken) {
      await this.login();
    }
  
    const fetchOptions = {
      method: "GET" as const,
      headers: {
        ...this.headers,
        auth_token: this.authToken!,
      },
      signal: AbortSignal.timeout(REQUEST_TIMEOUT_MS),
    };
  
    let response = await fetch(`${this.baseUrl}${path}`, fetchOptions);
  
    // Token may have expired — re-authenticate once and retry
    if (response.status === 401) {
      await this.login();
      response = await fetch(`${this.baseUrl}${path}`, {
        ...fetchOptions,
        headers: {
          ...this.headers,
          auth_token: this.authToken!,
        },
        signal: AbortSignal.timeout(REQUEST_TIMEOUT_MS),
      });
    }
  
    if (!response.ok) {
      const body = await response.text();
      throw new Error(
        `RAS API error (HTTP ${response.status}) on ${path}: ${body.substring(0, 300)}`
      );
    }
  
    return response.json();
  }