MCP Vision Relay

Instructions

Implementation Reference

• src/tools/geminiImage.ts:61-128 (handler)

  The execute function implementing the core logic of the gemini_analyze_image tool: prepares image, constructs prompt, calls runGeminiImageAnalysis, formats output with metadata, handles errors, and cleans up.
  async execute(args: GeminiArgs) { const prepared = await prepareImage(args.image, appConfig); const prompt = args.prompt?.trim() ?? appConfig.gemini.defaultPrompt ?? "Describe this image."; const attachment = createAttachmentReference(prepared.path); try { const result = await runGeminiImageAnalysis({ prompt, model: args.model, sandbox: args.sandbox, outputFormat: args.outputFormat, extraFlags: args.extraFlags, timeoutMs: args.timeoutMs ?? appConfig.commandTimeoutMs, imageReference: attachment, preparedImage: prepared, commandOverride: args.cliPath, }); const cleaned = result.stdout || "(Gemini CLI returned no output)"; const metaLines = [ `model: ${result.model ?? "default"}`, `imageSource: ${prepared.source}`, `durationMs: ${result.durationMs}`, ]; if (prepared.source === "local") { metaLines.push(`imagePath: ${prepared.path}`); } return { content: [ { type: "text", text: [ "### Gemini Output", cleaned, "", metaLines.join("\n"), ].join("\n"), }, ], isError: false, } as const; } catch (error) { if (error instanceof CommandError) { const details = [ error.message, error.result.stderr.trim() && `stderr:\n${error.result.stderr.trim()}`, error.result.stdout.trim() && `stdout:\n${error.result.stdout.trim()}`, ] .filter(Boolean) .join("\n\n"); return { content: [ { type: "text", text: `Gemini CLI failed:\n${details}`, }, ], isError: true, }; } throw error; } finally { await prepared.cleanup(); } },
• src/tools/geminiImage.ts:9-52 (schema)

  Zod schema defining the input parameters and validation for the gemini_analyze_image tool.
  const geminiSchema = z .object({ image: z .string() .min(1) .describe( "Local file path, http(s) URL, or base64-encoded image to analyze." ), prompt: z .string() .min(1) .optional() .describe("Instruction for Gemini. Defaults to config value."), model: z .string() .min(1) .optional() .describe("Gemini model identifier (e.g., gemini-2.0-flash)."), cliPath: z .string() .min(1) .optional() .describe("Override the Gemini CLI executable path."), sandbox: z .boolean() .optional() .describe("Whether to run the CLI with the sandbox flag (-s)."), outputFormat: z .enum(["text", "json"]) .optional() .describe("Request Gemini CLI to return the specified output format."), extraFlags: z .array(z.string().min(1)) .optional() .describe("Additional CLI flags to append as-is."), timeoutMs: z .number() .int() .positive() .max(600_000) .optional() .describe("Maximum time (in milliseconds) to wait for CLI execution."), }) .describe("Invoke Gemini CLI to analyze an image.");
• src/tools/geminiImage.ts:56-129 (registration)

  Tool registration call that defines the name, description, schema, and execute handler for gemini_analyze_image.
  registerTool({ name: "gemini_analyze_image", description: "Use Google Gemini CLI to describe or analyze an image using multimodal capabilities.", schema: geminiSchema, async execute(args: GeminiArgs) { const prepared = await prepareImage(args.image, appConfig); const prompt = args.prompt?.trim() ?? appConfig.gemini.defaultPrompt ?? "Describe this image."; const attachment = createAttachmentReference(prepared.path); try { const result = await runGeminiImageAnalysis({ prompt, model: args.model, sandbox: args.sandbox, outputFormat: args.outputFormat, extraFlags: args.extraFlags, timeoutMs: args.timeoutMs ?? appConfig.commandTimeoutMs, imageReference: attachment, preparedImage: prepared, commandOverride: args.cliPath, }); const cleaned = result.stdout || "(Gemini CLI returned no output)"; const metaLines = [ `model: ${result.model ?? "default"}`, `imageSource: ${prepared.source}`, `durationMs: ${result.durationMs}`, ]; if (prepared.source === "local") { metaLines.push(`imagePath: ${prepared.path}`); } return { content: [ { type: "text", text: [ "### Gemini Output", cleaned, "", metaLines.join("\n"), ].join("\n"), }, ], isError: false, } as const; } catch (error) { if (error instanceof CommandError) { const details = [ error.message, error.result.stderr.trim() && `stderr:\n${error.result.stderr.trim()}`, error.result.stdout.trim() && `stdout:\n${error.result.stdout.trim()}`, ] .filter(Boolean) .join("\n\n"); return { content: [ { type: "text", text: `Gemini CLI failed:\n${details}`, }, ], isError: true, }; } throw error; } finally { await prepared.cleanup(); } }, });
• src/providers/gemini.ts:25-67 (helper)

  Helper function that constructs and executes the Gemini CLI command for image analysis, handling arguments, prompt with image reference, and returning stdout and metadata.
  export async function runGeminiImageAnalysis( options: GeminiImageOptions ): Promise<GeminiSuccessResult> { const cliArgs: string[] = []; const model = options.model ?? appConfig.gemini.defaultModel; if (model) { cliArgs.push("-m", model); } const outputFormat = options.outputFormat ?? appConfig.gemini.defaultOutputFormat; if (outputFormat && outputFormat !== "text") { cliArgs.push("-o", outputFormat); } if (options.sandbox) { cliArgs.push("-s"); } if (appConfig.gemini.extraArgs?.length) { cliArgs.push(...appConfig.gemini.extraArgs); } if (options.extraFlags?.length) { cliArgs.push(...options.extraFlags); } const finalPrompt = `${options.prompt.trim()}\n\n${options.imageReference}`; cliArgs.push("-p", finalPrompt); const command = options.commandOverride ?? appConfig.gemini.command; const result = await runCommand(command, cliArgs, { timeoutMs: options.timeoutMs, }); return { stdout: result.stdout.trim(), durationMs: result.durationMs, model, }; }