fix: prevent memory explosion in search_files_content#2983
Merged
Conversation
Use stat (following symlinks) instead of lstat when checking file size, so symlinks to large files are skipped rather than read whole. Also skip non-regular files to avoid reading forever from devices like /dev/zero. Remove unreachable early return of fs.SkipAll.
gtardif
approved these changes
Jun 3, 2026
docker-agent
reviewed
Jun 3, 2026
docker-agent
left a comment
docker-agent
left a comment
There was a problem hiding this comment.
Assessment: 🟢 APPROVE
The changes are correct and well-implemented. The output cap (1 MiB), large-file skip (10 MiB), symlink-aware stat guard, and incremental strings.Builder pattern all work as intended. Tests cover truncation, large-file skipping, and the symlink-to-large-file regression case.
aheritier
added
area/tools
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The
search_files_contentfilesystem tool can produce massive results when searching across large directory trees, causing unbounded memory growth as the output cascades through multiple layers: tool output buffer → in-memory message list → session store/WAL → re-serialization into every subsequent model request for the rest of the session.This change caps the total output at 1 MiB with a truncation notice and skips files larger than 10 MiB. Output is now built incrementally with
strings.Builderinstead of accumulating a slice and joining at the end. The file-size check now stats the target (following symlinks, not the symlink itself) and only scans regular files, which also prevents reading forever from device files or FIFOs like/dev/zero.Tests cover output-cap truncation, large-file skipping, and the symlink-to-large-file case.