From 1ded0d375920e82364ea507ea1c28da02b6eb7fb Mon Sep 17 00:00:00 2001
From: can1357 <me@can.ac>
Date: Thu, 2 Jul 2026 02:40:07 +0200
Subject: [PATCH] fix(coding-agent): restricted key extraction to top-level
 json depth

- Introduced JSON depth tracking to ensure string keys are only extracted at the top level.
- Prevented nested keys matching designated streaming keys from being prematurely or incorrectly captured.
- Added comprehensive unit tests validating nested key exclusion and correct top-level extraction order.
---
 .../src/modes/controllers/tool-args-reveal.ts | 11 +++++++-
 .../test/tool-args-reveal.test.ts             | 25 +++++++++++++++++++
 2 files changed, 35 insertions(+), 1 deletion(-)

diff --git a/packages/coding-agent/src/modes/controllers/tool-args-reveal.ts b/packages/coding-agent/src/modes/controllers/tool-args-reveal.ts
index aa412f3ac..a4c67905e 100644
--- a/packages/coding-agent/src/modes/controllers/tool-args-reveal.ts
+++ b/packages/coding-agent/src/modes/controllers/tool-args-reveal.ts
@@ -64,6 +64,10 @@ class StreamingJsonStringExtractor {
 	readonly #keys: Set<string>;
 	#source = "";
 	#offset = 0;
+	/** `{`/`[` nesting outside strings. Candidate keys match only at depth 1 —
+	 *  the top level of the args object — so a nested object's key (e.g.
+	 *  `{"meta":{"content":…}}`) is never captured as a streamed top-level arg. */
+	#depth = 0;
 	#state: StreamingJsonStringExtractorState = "scan";
 	#candidate = "";
 	#candidateEscaped = false;
@@ -82,6 +86,7 @@ class StreamingJsonStringExtractor {
 	reset(): void {
 		this.#source = "";
 		this.#offset = 0;
+		this.#depth = 0;
 		this.#state = "scan";
 		this.#candidate = "";
 		this.#candidateEscaped = false;
@@ -129,6 +134,10 @@ class StreamingJsonStringExtractor {
 			this.#candidateEscaped = false;
 			this.#candidateUnicode = "";
 			this.#state = "candidate";
+		} else if (ch === "{" || ch === "[") {
+			this.#depth++;
+		} else if (ch === "}" || ch === "]") {
+			this.#depth--;
 		}
 		this.#offset++;
 	}
@@ -154,7 +163,7 @@ class StreamingJsonStringExtractor {
 			return;
 		}
 		if (ch === '"') {
-			this.#matchedKey = this.#keys.has(this.#candidate) ? this.#candidate : undefined;
+			this.#matchedKey = this.#depth === 1 && this.#keys.has(this.#candidate) ? this.#candidate : undefined;
 			this.#state = "afterCandidate";
 			this.#offset++;
 			return;
diff --git a/packages/coding-agent/test/tool-args-reveal.test.ts b/packages/coding-agent/test/tool-args-reveal.test.ts
index 7dbe7abd4..651c709f7 100644
--- a/packages/coding-agent/test/tool-args-reveal.test.ts
+++ b/packages/coding-agent/test/tool-args-reveal.test.ts
@@ -193,6 +193,31 @@ describe("tool args reveal", () => {
 		expect(latest!.content).toBe(`${initialContent}${appendedContent}`);
 	});
 
+	it("ignores streaming string keys nested below the top level", () => {
+		vi.useFakeTimers();
+		const { controller } = makeController({ smooth: false });
+		const streamingStringKeys = ["content"];
+
+		// Prefix ends inside the nested object: the nested "content" key must
+		// NOT be captured and injected as a top-level preview arg.
+		const nestedOnly = `{"meta":{"content":"NESTED"`;
+		const first = controller.setTarget("call-1", nestedOnly, jsonTarget({ streamingStringKeys }));
+		expect(first.content).toBeUndefined();
+
+		// Growth past the nested object: the real top-level key must be captured.
+		const full = `${nestedOnly}"},"content":"real"}`;
+		const second = controller.setTarget("call-1", full, jsonTarget({ streamingStringKeys }));
+		expect(second.content).toBe("real");
+
+		// Reverse order: a nested duplicate arriving AFTER the top-level key must
+		// not reset/overwrite the captured top-level value. Extractor values are
+		// merged over the parsed args, so a depth-blind capture would win here.
+		const reversed = `{"content":"real","meta":{"content":"NESTED"}}`;
+		const decoded = decodeStreamedToolArgs(reversed, { rawInput: false, streamingStringKeys });
+		expect(decoded.content).toBe("real");
+		expect(decoded.meta).toEqual({ content: "NESTED" });
+	});
+
 	it("decodes the full received buffer, unpaced, when smoothing is disabled", () => {
 		vi.useFakeTimers();
 		const requestRender = vi.fn();