From 8676a0dcd8f9f6bca78835e63c8cd31493c4154d Mon Sep 17 00:00:00 2001
From: Armin Ronacher <armin.ronacher@active-4.com>
Date: Thu, 24 Sep 2026 09:15:58 +0200
Subject: [PATCH] fix(coding-agent): require advertised X11 clipboard images

closes #9786
---
 packages/coding-agent/CHANGELOG.md            |  1 +
 .../coding-agent/src/utils/clipboard-image.ts | 29 +++++++----------
 .../coding-agent/test/clipboard-image.test.ts | 31 +++++++++++++++++--
 3 files changed, 41 insertions(+), 20 deletions(-)

diff --git a/packages/coding-agent/CHANGELOG.md b/packages/coding-agent/CHANGELOG.md
index 8856920fe..1a965f3b1 100644
--- a/packages/coding-agent/CHANGELOG.md
+++ b/packages/coding-agent/CHANGELOG.md
@@ -11,6 +11,7 @@
 
 ### Fixed
 
+- Fixed X11 clipboard text being misidentified as an image when the clipboard owner accepts unadvertised image targets ([#9786](https://github.com/earendil-works/pi/issues/9786)).
 - Prevented managed git packages from automatically installing Pi peer dependencies and added warnings for extension packages that list host-provided modules in `dependencies` ([#9863](https://github.com/earendil-works/pi/issues/9863)).
 
 ## [0.87.1] - 2026-09-22
diff --git a/packages/coding-agent/src/utils/clipboard-image.ts b/packages/coding-agent/src/utils/clipboard-image.ts
index 086f1fadf..f6e7ecd54 100644
--- a/packages/coding-agent/src/utils/clipboard-image.ts
+++ b/packages/coding-agent/src/utils/clipboard-image.ts
@@ -174,27 +174,20 @@ async function readClipboardImageViaXclip(): Promise<ClipboardImage | null | und
 		timeoutMs: DEFAULT_LIST_TIMEOUT_MS,
 	});
 
-	let candidateTypes: string[] = [];
-	if (targets !== undefined) {
-		candidateTypes = targets
-			.toString("utf-8")
-			.split(/\r?\n/)
-			.map((t) => t.trim())
-			.filter(Boolean);
-	}
+	if (targets === undefined) return undefined;
 
+	const candidateTypes = targets
+		.toString("utf-8")
+		.split(/\r?\n/)
+		.map((t) => t.trim())
+		.filter(Boolean);
 	const preferred = selectPreferredImageMimeType(candidateTypes);
-	if (targets !== undefined && !preferred) return null;
-	const tryTypes = new Set(preferred ? [preferred, ...SUPPORTED_IMAGE_MIME_TYPES] : SUPPORTED_IMAGE_MIME_TYPES);
+	if (!preferred) return null;
 
-	for (const mimeType of tryTypes) {
-		const data = await runClipboardCommand("xclip", ["-selection", "clipboard", "-t", mimeType, "-o"]);
-		if (data !== undefined && data.length > 0) {
-			return { bytes: data, mimeType: baseMimeType(mimeType) };
-		}
-	}
-
-	return undefined;
+	const data = await runClipboardCommand("xclip", ["-selection", "clipboard", "-t", preferred, "-o"]);
+	if (data === undefined) return undefined;
+	if (data.length === 0) return null;
+	return { bytes: data, mimeType: baseMimeType(preferred) };
 }
 
 async function readClipboardImageViaNativeClipboard(): Promise<ClipboardImage | null | undefined> {
diff --git a/packages/coding-agent/test/clipboard-image.test.ts b/packages/coding-agent/test/clipboard-image.test.ts
index e7aacc4fb..89319142d 100644
--- a/packages/coding-agent/test/clipboard-image.test.ts
+++ b/packages/coding-agent/test/clipboard-image.test.ts
@@ -44,6 +44,33 @@ describe("readClipboardImage", () => {
 		});
 	}
 
+	test("X11 does not probe image types when TARGETS fails", async () => {
+		// Regression test for #9786.
+		mocks.getImage.mockResolvedValue(null);
+		mocks.command.mockImplementation(async (_command, args) =>
+			args.includes("TARGETS") ? undefined : Buffer.from("hello"),
+		);
+		expect(await readClipboardImage({ platform: "linux", env: { DISPLAY: ":0" } })).toBeNull();
+		expect(mocks.command).toHaveBeenCalledExactlyOnceWith(
+			"xclip",
+			["-selection", "clipboard", "-t", "TARGETS", "-o"],
+			{ timeoutMs: 1000 },
+		);
+		expect(mocks.getImage).toHaveBeenCalledOnce();
+	});
+
+	test("X11 does not probe unadvertised image types", async () => {
+		mocks.getImage.mockResolvedValue(null);
+		mocks.command.mockImplementation(async (_command, args) => {
+			if (args.includes("TARGETS")) return Buffer.from("image/png\n");
+			if (args.includes("image/png")) return undefined;
+			return Buffer.from("hello");
+		});
+		expect(await readClipboardImage({ platform: "linux", env: { DISPLAY: ":0" } })).toBeNull();
+		expect(mocks.command.mock.calls.map(([, args]) => args[3])).toEqual(["TARGETS", "image/png"]);
+		expect(mocks.getImage).toHaveBeenCalledOnce();
+	});
+
 	test.each([png, null, new Uint8Array()])("native X11 result %j stops fallback", async (bytes) => {
 		mocks.getImage.mockResolvedValue(bytes);
 		expect(await readClipboardImage({ platform: "linux", env: { DISPLAY: ":0" } })).toEqual(
@@ -51,7 +78,7 @@ describe("readClipboardImage", () => {
 		);
 		expect(mocks.getNativeClipboard).toHaveBeenCalledExactlyOnceWith();
 		expect(mocks.getImage).toHaveBeenCalledOnce();
-		expect(mocks.command.mock.calls.map(([name]) => name)).toEqual(Array<string>(5).fill("xclip"));
+		expect(mocks.command.mock.calls.map(([name]) => name)).toEqual(["xclip"]);
 	});
 	test.each(["missing module", "unavailable display"])("Wayland: falls back to X11 after %s", async (failure) => {
 		if (failure === "missing module") mocks.getNativeClipboard.mockReturnValue(undefined);
@@ -121,7 +148,7 @@ describe("readClipboardImage", () => {
 				error,
 			);
 			expect(mocks.command.mock.calls.map(([name]) => name)).toEqual(
-				platform === "linux" ? ["wl-paste", ...Array<string>(5).fill("xclip")] : [],
+				platform === "linux" ? ["wl-paste", "xclip"] : [],
 			);
 		},
 	);