pi-banana 2.1.4 → 2.4.0

package/README.md

@@ -58,7 +58,7 @@ The model calls `banana_image` or `banana_vision` automatically.
 | `prompt` | string | — | Required. What to draw, or what to change about the reference image. |
 | `aspectRatio` | enum | `1:1` | `1:1`, `2:3`, `3:2`, `3:4`, `4:3`, `4:5`, `5:4`, `9:16`, `16:9`, `21:9` |
 | `imageSize` | enum | `1K` | `1K`, `2K`, `4K`. `4K` requires `quality=high`. |
-| `quality` | enum | `fast` | `fast` = Nano Banana 2 (~3–10 s, cheapest). `high` = Nano Banana Pro (slower, top quality). |
+| `quality` | enum | `fast` | `lite` = Nano Banana 2 Lite (ultra-low latency, cheapest). `fast` = Nano Banana 2 (~3–10 s). `high` = Nano Banana Pro (slower, top quality). All GA model ids since 2.4.0. |
 | `referenceImages` | array | — | Optional array of paths (PNG/JPEG/WebP/GIF) to edit or use for composition instead of generating from scratch. |
 | `outputPath` | string | `./generated/<slug>-<ts>.png` | Optional output path. Parent dirs are created. |
 

package/extensions/index.ts

@@ -9,12 +9,16 @@
  * pi-coding-agent renders them inline (Kitty / iTerm2). Also writes the file
  * to disk so the agent and the user always have a path to refer back to.
  *
- * Auth: GOOGLE_API_KEY in the environment.
- *   - AI Studio keys (AIza...) hit the Gemini API.
- *   - Vertex AI Express keys (AQ....) automatically switch to Vertex.
+ * Auth (checked in this order — first match wins):
+ *   1. pi modelRegistry → auth.json / env vars / models.json
+ *      (set via /login → Google, or GEMINI_API_KEY env var)
+ *   2. Vertex AI ADC — gcloud auth application-default login,
+ *      GOOGLE_APPLICATION_CREDENTIALS, or GOOGLE_CLOUD_PROJECT
+ *   3. GOOGLE_API_KEY env var (AIza… or AQ.…) — pi-banana backward compat
+ *   Vertex AI Express keys (AQ.…) automatically switch to Vertex.
  */
 
-import type { ExtensionAPI } from "@earendil-works/pi-coding-agent";
+import type { ExtensionAPI, ExtensionContext } from "@earendil-works/pi-coding-agent";
 import { StringEnum } from "@earendil-works/pi-ai";
 import {
 	Box,
@@ -25,6 +29,7 @@ import {
 } from "@earendil-works/pi-tui";
 import { GoogleGenAI } from "@google/genai";
 import { existsSync } from "node:fs";
+import { homedir } from "node:os";
 import { mkdir, readFile, writeFile } from "node:fs/promises";
 import { extname, isAbsolute, resolve } from "node:path";
 import { Type } from "typebox";
@@ -46,26 +51,26 @@ const ASPECT_RATIOS = [
 	"21:9",
 ] as const;
 const IMAGE_SIZES = ["1K", "2K", "4K"] as const;
-const QUALITY = ["fast", "high"] as const;
-
-// Resolved against a live Vertex Express (AQ.*) key on 2026-05-08:
-//   gemini-3.1-flash-image          → 404
-//   gemini-3.1-flash-image-preview  → ok  (Nano Banana 2)
-//   gemini-3-pro-image              → 404
-//   gemini-3-pro-image-preview      → ok  (Nano Banana Pro)
-// AI Studio keys (AIza*) sometimes publish the un-suffixed GA names.
-// `callWithModelFallback` below transparently retries with `-preview`
-// if the un-suffixed id 404s, so both backends just work.
-const MODEL_FOR_QUALITY: Record<(typeof QUALITY)[number], string> = {
-	fast: "gemini-3.1-flash-image-preview", // Nano Banana 2
-	high: "gemini-3-pro-image-preview", // Nano Banana Pro
+const IMAGE_QUALITY = ["lite", "fast", "high"] as const;
+const VISION_QUALITY = ["fast", "high"] as const;
+type ImageQuality = (typeof IMAGE_QUALITY)[number];
+
+// GA ids (Gemini API changelog 2026-05-28: gemini-3.1-flash-image and
+// gemini-3-pro-image went GA; 2026-06-30: gemini-3.1-flash-lite-image GA).
+// Vertex Express keys historically served the `-preview` variants; the
+// GA and preview ids are published at different times per backend, so
+// `callWithModelFallback` retries the other variant on 404. Both work.
+// Verified against the Gemini API changelog on 2026-09-01.
+const MODEL_FOR_QUALITY: Record<ImageQuality, string> = {
+	lite: "gemini-3.1-flash-lite-image", // Nano Banana 2 Lite (ultra-low latency, cheapest)
+	fast: "gemini-3.1-flash-image", // Nano Banana 2
+	high: "gemini-3-pro-image", // Nano Banana Pro
 };
 
 const DEFAULT_OUTPUT_DIR =
 	process.env.PI_IMAGE_DIR?.trim() || "generated";
 const DEFAULT_QUALITY = (process.env.PI_IMAGE_QUALITY as
-	| "fast"
-	| "high"
+	| ImageQuality
 	| undefined) ?? "fast";
 const SUPPORTED_INPUT_MIME = new Set([
 	"image/png",
@@ -76,19 +81,71 @@ const SUPPORTED_INPUT_MIME = new Set([
 
 // ─── Helpers ───────────────────────────────────────────────────────────────
 
-function buildClient(): GoogleGenAI {
-	const apiKey = process.env.GOOGLE_API_KEY;
-	if (!apiKey) {
-		throw new Error(
-			"GOOGLE_API_KEY not set. Get one at https://aistudio.google.com/apikey (AIza…) " +
-				"or use a Vertex AI Express key (AQ.…) — both work.",
-		);
+// Well-known path for gcloud Application Default Credentials.
+const ADC_CREDENTIALS_PATH = resolve(
+	homedir(),
+	".config/gcloud/application_default_credentials.json",
+);
+
+/**
+ * Check whether Vertex AI Application Default Credentials are available.
+ * Returns true when any of these are configured:
+ *   - GOOGLE_APPLICATION_CREDENTIALS points to an existing file
+ *   - gcloud ADC file exists (~/.config/gcloud/application_default_credentials.json)
+ *   - GOOGLE_CLOUD_PROJECT is set (implies GKE / Cloud Run / Compute Engine default SA)
+ */
+function hasVertexADC(): boolean {
+	const gac = process.env.GOOGLE_APPLICATION_CREDENTIALS;
+	if (gac && existsSync(gac)) return true;
+	if (existsSync(ADC_CREDENTIALS_PATH)) return true;
+	if (process.env.GOOGLE_CLOUD_PROJECT) return true;
+	return false;
+}
+
+async function buildClient(
+	modelRegistry: ExtensionContext["modelRegistry"] | undefined,
+): Promise<GoogleGenAI> {
+	// 1. Pi's model registry — the pi-native auth chain
+	//    (auth.json → env vars → models.json fallback)
+	if (modelRegistry) {
+		try {
+			const apiKey = await modelRegistry.getApiKeyForProvider("google");
+			if (apiKey) {
+				return new GoogleGenAI({
+					apiKey,
+					...(apiKey.startsWith("AQ.") ? { vertexai: true } : {}),
+				});
+			}
+		} catch {
+			// ignore — fall through
+		}
 	}
-	// AQ.* express keys authenticate against Vertex; AIza* against Gemini API.
-	return new GoogleGenAI({
-		apiKey,
-		...(apiKey.startsWith("AQ.") ? { vertexai: true } : {}),
-	});
+
+	// 2. Vertex AI ADC (no API key needed — SDK picks up ADC automatically)
+	if (hasVertexADC()) {
+		return new GoogleGenAI({ vertexai: true });
+	}
+
+	// 3. GOOGLE_API_KEY env var — backward compat for pi-banana users
+	//    (pi's standard GEMINI_API_KEY is already covered by modelRegistry above)
+	if (process.env.GOOGLE_API_KEY) {
+		const key = process.env.GOOGLE_API_KEY;
+		return new GoogleGenAI({
+			apiKey: key,
+			...(key.startsWith("AQ.") ? { vertexai: true } : {}),
+		});
+	}
+
+	throw new Error(
+		"No Google credentials found. pi-banana looked in this order:\n" +
+			"  1. pi model registry (auth.json → env vars → models.json)\n" +
+			"     Run /login → Google, or set GEMINI_API_KEY, or configure in models.json\n" +
+			"  2. Vertex AI ADC (gcloud auth application-default login,\n" +
+			"     GOOGLE_APPLICATION_CREDENTIALS, or GOOGLE_CLOUD_PROJECT)\n" +
+			"  3. GOOGLE_API_KEY env var (AIza… or AQ.…)\n\n" +
+			"For Gemini API: get a key at https://aistudio.google.com/apikey\n" +
+			"For Vertex AI: gcloud auth application-default login",
+	);
 }
 
 function slugify(text: string): string {
@@ -193,7 +250,7 @@ export default function (pi: ExtensionAPI) {
 			"inline in the terminal AND saved to disk so it can be re-used. " +
 			"Pass `referenceImages` (array of paths) to iterate on existing pictures.",
 		promptSnippet:
-			"Generate or edit images with Google Nano Banana via the GOOGLE_API_KEY env var.",
+			"Generate or edit images with Google Nano Banana. Requires Google credentials (via /login, GOOGLE_API_KEY / GEMINI_API_KEY, or Vertex AI ADC).",
 		promptGuidelines: [
 			"Call banana_image when the user asks to create, draw, illustrate, or edit a picture.",
 			"For tweaks like 'make the sky purple', pass the previous file path (or paths) as referenceImages to banana_image.",
@@ -220,9 +277,9 @@ export default function (pi: ExtensionAPI) {
 				}),
 			),
 			quality: Type.Optional(
-				StringEnum(QUALITY, {
+				StringEnum(IMAGE_QUALITY, {
 					description:
-						"'fast' = Nano Banana 2 (default, ~3s, cheap). 'high' = Nano Banana Pro (slower, top quality).",
+						"'lite' = Nano Banana 2 Lite (ultra-low latency, cheapest). 'fast' = Nano Banana 2 (default, ~3s, cheap). 'high' = Nano Banana Pro (slower, top quality).",
 					default: DEFAULT_QUALITY,
 				}),
 			),
@@ -269,7 +326,7 @@ export default function (pi: ExtensionAPI) {
 				);
 			}
 
-			const client = buildClient();
+			const client = await buildClient(ctx.modelRegistry);
 
 			// Build content parts — text + (optional) reference image.
 			const parts: Array<{
@@ -291,7 +348,7 @@ export default function (pi: ExtensionAPI) {
 				content: [
 					{
 						type: "text",
-						text: `🎨 ${verbLabel} ${aspectRatio} ${imageSize} image with ${quality === "fast" ? "Nano Banana 2" : "Nano Banana Pro"}…`,
+						text: `🎨 ${verbLabel} ${aspectRatio} ${imageSize} image with ${quality === "lite" ? "Nano Banana 2 Lite" : quality === "fast" ? "Nano Banana 2" : "Nano Banana Pro"}…`,
 					},
 				],
 				details: { model, aspectRatio, imageSize, quality, editing },
@@ -422,7 +479,6 @@ export default function (pi: ExtensionAPI) {
 				container.addChild(new Spacer(1));
 				container.addChild(
 					new Image((details as any).imageBase64, (details as any).mimeType ?? "image/png", {
-						...theme,
 						fallbackColor: (s: string) => theme.fg("muted", s),
 					}, {
 						maxWidthCells: 80,
@@ -503,10 +559,10 @@ export default function (pi: ExtensionAPI) {
 					"to analyze (e.g., ['image1.png']). Relative paths resolve to current working directory.",
 			}),
 			quality: Type.Optional(
-				StringEnum(QUALITY, {
+				StringEnum(VISION_QUALITY, {
 					description:
 						"'fast' = gemini-3.1-flash-lite (default, fast/cheap). 'high' = gemini-3.1-pro-preview (slower, deep reasoning).",
-					default: DEFAULT_QUALITY,
+					default: "fast",
 				}),
 			),
 		}),
@@ -528,7 +584,7 @@ export default function (pi: ExtensionAPI) {
 				return { content: [{ type: "text", text: "Cancelled." }], details: {} };
 			}
 
-			const client = buildClient();
+			const client = await buildClient(ctx.modelRegistry);
 			const parts: Array<{
 				text?: string;
 				inlineData?: { mimeType: string; data: string };

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "pi-banana",
-  "version": "2.1.4",
+  "version": "2.4.0",
   "description": "Generate, edit, and analyze images in pi using Google Nano Banana (image gen) and Gemini Vision (analysis). Inline terminal preview, reference-image editing, auto-save.",
   "author": "Francesco Frapporti <effedue@gmail.com>",
   "license": "MIT",