Agents unable to read long file in its entirety

[jojo2357]

I am using Qwen 3.5 9B and I have noticed that for large files (the cutoff is around 500-1000 lines?) the read_file truncates the file and my agents get stuck in a loop just retrying to read the same file over and over.

How can i change this behavior? is the read function itself truncating the file contents?

my model settings in qwen.json are:

"generationConfig": {
  "timeout": 300000,
  "maxRetries": 1,
  "contextWindowSize": 262144,
  "samplingParams": {
    "temperature": 0.6,
    "top_p": 0.95,
    "top_k": 20,
    "min_p": 0,
    "presence_penalty": 0,
    "repetition_penalty": 1,
    "max_tokens": 32768
  }
}

[Maxaubert]

Yes, it is read_file itself truncating, and it is configurable. It is not related to your generationConfig or the model's context window.

Cause: read_file applies two caps, both taken from the tools settings:

• tools.truncateToolOutputLines, default 1000 lines (used when the model does not pass its own limit)
• tools.truncateToolOutputThreshold, default 25000 characters (always applied)

The char cap is the one you most likely hit at around 500 to 1000 lines, since 25k chars is only about 500 lines of typical code. When either cap trips, the output is cut and marked as truncated, and the model is supposed to continue with offset and limit. A 9B model often doesn't, so it re-reads the same first chunk in a loop.

Defaults:

qwen-code/packages/core/src/config/config.ts

Lines 877 to 878 in 9f6138a

	export const DEFAULT_TRUNCATE_TOOL_OUTPUT_THRESHOLD = 25_000;
	export const DEFAULT_TRUNCATE_TOOL_OUTPUT_LINES = 1000;

Where they are applied in read_file:

qwen-code/packages/core/src/utils/fileUtils.ts

Lines 1512 to 1580 in 9f6138a

	// Use BOM-aware reader to avoid leaving a BOM character in content and to support UTF-16/32 transparently
	const fileSystemService = config.getFileSystemService();
	const maxOutputBytes = getRangeReadByteLimit(config);
	const readTextFileFromHandle = fileSystemService.readTextFileFromHandle;
	const { content, _meta } = options.textFileHandle
	? await readTextFileFromHandle!.call(fileSystemService, {
	fileHandle: options.textFileHandle,
	fileSize: stats.size,
	limit: limit ?? config.getTruncateToolOutputLines(),
	line: offset,
	maxOutputBytes,
	maxScanBytes: options.textFileMaxScanBytes ?? maxOutputBytes,
	...(signal !== undefined ? { signal } : {}),
	})
	: await fileSystemService.readTextFile({
	path: filePath,
	limit: limit ?? config.getTruncateToolOutputLines(),
	line: offset,
	maxOutputBytes,
	stats,
	...(signal !== undefined ? { signal } : {}),
	});
	const selectedLines = content.split('\n').map((line) => line.trimEnd());
	const startLine = offset || 0;
	const selectedLineCount =
	content.length === 0 ? 0 : selectedLines.length;
	const hasOriginalLineCount = _meta?.originalLineCount !== undefined;
	const originalLineCount =
	_meta?.originalLineCount ??
	(stats.size >= TEXT_RANGE_FAST_PATH_MAX_SIZE
	? startLine + selectedLineCount
	: await countFileLines(filePath));
	const originalLineCountExact =
	_meta?.originalLineCountExact === false
	? false
	: hasOriginalLineCount ||
	stats.size < TEXT_RANGE_FAST_PATH_MAX_SIZE;
	const configCharLimit = config.getTruncateToolOutputThreshold();
	// Apply character limit truncation
	let llmContent = '';
	let contentLengthTruncated = false;
	let linesIncluded = 0;
	if (Number.isFinite(configCharLimit)) {
	const formattedLines: string[] = [];
	let currentLength = 0;
	for (const line of selectedLines) {
	const sep = linesIncluded > 0 ? 1 : 0; // newline separator
	linesIncluded++;
	const projectedLength = currentLength + line.length + sep;
	if (projectedLength <= configCharLimit) {
	formattedLines.push(line);
	currentLength = projectedLength;
	} else {
	// Truncate the current line to fit
	const remaining = Math.max(
	configCharLimit - currentLength - sep,
	10,
	);
	formattedLines.push(
	line.substring(0, remaining) + '... [truncated]',
	);
	contentLengthTruncated = true;
	break;
	}
	}

A value of 0 or less means no limit (it is mapped to Infinity):

qwen-code/packages/core/src/config/config.ts

Lines 10757 to 10775 in 9f6138a

	getTruncateToolOutputThreshold(): number {
	if (this.truncateToolOutputThreshold <= 0) {
	return Number.POSITIVE_INFINITY;
	}
	return this.truncateToolOutputThreshold;
	}
	isTruncateToolOutputThresholdExplicit(): boolean {
	return this.truncateToolOutputThresholdExplicit;
	}
	getTruncateToolOutputLines(): number {
	if (this.truncateToolOutputLines <= 0) {
	return Number.POSITIVE_INFINITY;
	}
	return this.truncateToolOutputLines;
	}

Fix: raise them in ~/.qwen/settings.json (or .qwen/settings.json in the project). These are not part of qwen.json model settings, and they need a restart:

{
  "tools": {
    "truncateToolOutputThreshold": 200000,
    "truncateToolOutputLines": 5000
  }
}

Docs:

qwen-code/docs/users/configuration/settings.md

Lines 402 to 403 in 9f6138a

	| `tools.truncateToolOutputThreshold` | number | Truncate tool output if it is larger than this many characters. Applies to Shell, Grep, Glob, ReadFile and ReadManyFiles tools. | `25000` | Requires restart: Yes |
	| `tools.truncateToolOutputLines` | number | Maximum lines or entries kept when truncating tool output. Applies to Shell, Grep, Glob, ReadFile and ReadManyFiles tools. | `1000` | Requires restart: Yes |

Keep in mind these settings also apply to Shell, Grep, Glob and ReadManyFiles output, and a big file read eats context (a 262k window is fine for a few thousand lines, but not unlimited). Setting them to a moderate value rather than 0 is safer.

[jojo2357]

Thank you so much for your answer, I just could NOT find it on my own