pi-banana 2.1.2 → 2.3.0

package/extensions/index.ts

@@ -9,9 +9,13 @@
  * 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";
@@ -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";
@@ -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: ExtensionAPI["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.",
@@ -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<{
@@ -385,12 +442,10 @@ export default function (pi: ExtensionAPI) {
 				: `Generated → ${outPath}`;
 
 			return {
-				// Flat ImageContent block — pi-coding-agent auto-renders inline
-				// in Kitty/iTerm2 terminals. Older terminals get the text fallback.
-				content: [
-					{ type: "text", text: summary },
-					{ type: "image", data: base64, mimeType },
-				],
+				// Only text in content — the base64 image goes into details
+				// so it does not bloat conversation history sent to the LLM.
+				// renderResult reads the image from details for inline display.
+				content: [{ type: "text", text: summary }],
 				details: {
 					prompt: params.prompt,
 					model,
@@ -401,25 +456,29 @@ export default function (pi: ExtensionAPI) {
 					outputPath: outPath,
 					editing,
 					referenceImages: params.referenceImages,
+					imageBase64: base64,
 				},
 			};
 		},
 
 		renderResult(result, _options, theme) {
-			// existing renderResult code...
+			// Image is in details (not content) so base64 does not bloat
+			// conversation history sent to the LLM on subsequent turns.
 			const { details, content } = result;
 			const container = new Container();
 
 			// 1. Summary
 			const summaryPart = content.find((c: any) => c.type === "text");
 			const summaryText = (summaryPart && summaryPart.type === "text") ? summaryPart.text : "";
-			container.addChild(new Text(theme.fg("success", "✔ " + summaryText), 1, 0));
+			if (summaryText) {
+				container.addChild(new Text(theme.fg("success", "✔ " + summaryText), 1, 0));
+			}
 
-			// 2. Image
-			const imagePart = content.find((c: any) => c.type === "image");
-			if (imagePart && imagePart.type === "image") {
+			// 2. Image from details
+			if (details && (details as any).imageBase64) {
+				container.addChild(new Spacer(1));
 				container.addChild(
-					new Image(imagePart.data, imagePart.mimeType, {
+					new Image((details as any).imageBase64, (details as any).mimeType ?? "image/png", {
 						...theme,
 						fallbackColor: (s: string) => theme.fg("muted", s),
 					}, {
@@ -526,7 +585,7 @@ 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": "2.1.2",
+  "version": "2.3.0",
   "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",