Skip to content

feat: make tool output size limit configurable via GOOSE_MAX_TOOL_RESPONSE_SIZE #9256

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
DOsinga wants to merge 1 commit into
base: main
Choose a base branch
from
+12 −4
Open

feat: make tool output size limit configurable via GOOSE_MAX_TOOL_RESPONSE_SIZE #9256

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
16 changes: 12 additions & 4 deletions crates/goose/src/agents/large_response_handler.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,14 +1,22 @@
use crate::config::Config;
use chrono::Utc;
use rmcp::model::{CallToolResult, Content, ErrorData};
use std::fs::File;
use std::io::Write;

const LARGE_TEXT_THRESHOLD: usize = 200_000;
const DEFAULT_LARGE_TEXT_THRESHOLD: usize = 200_000;

fn large_text_threshold() -> usize {
Config::global()
.get_param::<usize>("GOOSE_MAX_TOOL_RESPONSE_SIZE")
.unwrap_or(DEFAULT_LARGE_TEXT_THRESHOLD)
Comment on lines +9 to +12
Copy link
Copy Markdown

@chatgpt-codex-connector chatgpt-codex-connector Bot May 15, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P2 Badge Cache configured response-size limit

Avoid reading GOOSE_MAX_TOOL_RESPONSE_SIZE from Config on every tool result. process_tool_response runs for each tool call, and large_text_threshold() now calls Config::global().get_param(...), which falls through to config-file loading/parsing when the env var is unset. That introduces synchronous filesystem/config-deserialization work on a hot path and can add noticeable latency in tool-heavy sessions compared to the previous constant-time check.

Useful? React with 👍 / 👎.

}

/// Process tool response and handle large text content
pub fn process_tool_response(
response: Result<CallToolResult, ErrorData>,
) -> Result<CallToolResult, ErrorData> {
let threshold = large_text_threshold();
match response {
Ok(mut result) => {
let mut processed_contents = Vec::new();
Expand All @@ -17,7 +25,7 @@ pub fn process_tool_response(
match content.as_text() {
Some(text_content) => {
// Check if text exceeds threshold
if text_content.text.chars().count() > LARGE_TEXT_THRESHOLD {
if text_content.text.chars().count() > threshold {
// Write to temp file
match write_large_text_to_file(&text_content.text) {
Ok(file_path) => {
Expand Down Expand Up @@ -107,7 +115,7 @@ mod tests {
#[test]
fn test_large_text_response_redirected_to_file() {
// Create a text larger than the threshold
let large_text = "a".repeat(LARGE_TEXT_THRESHOLD + 1000);
let large_text = "a".repeat(DEFAULT_LARGE_TEXT_THRESHOLD + 1000);
let content = Content::text(large_text.clone());

let response = Ok(CallToolResult::success(vec![content]));
Expand Down Expand Up @@ -166,7 +174,7 @@ mod tests {
fn test_mixed_content_handled_correctly() {
// Create a response with mixed content types
let small_text = Content::text("Small text");
let large_text = Content::text("a".repeat(LARGE_TEXT_THRESHOLD + 1000));
let large_text = Content::text("a".repeat(DEFAULT_LARGE_TEXT_THRESHOLD + 1000));
let image = Content::image("image_data".to_string(), "image/jpeg".to_string());

let response = Ok(CallToolResult::success(vec![small_text, large_text, image]));
Expand Down
Loading