fix(browseros-agent): strip leaked model scaffold tokens from chat output#1782
Open
GriggsRice wants to merge 1 commit into
Open
fix(browseros-agent): strip leaked model scaffold tokens from chat output#1782GriggsRice wants to merge 1 commit into
GriggsRice wants to merge 1 commit into
Conversation
…tput
Some local models (observed via LM Studio) embed their own
thinking-scaffold markers, e.g. "<|channel|>" or the malformed
"<channel|>", directly in regular text/reasoning output instead of a
structured field. Nothing sanitized this before rendering, so raw
protocol tokens showed up verbatim in the chat UI.
Add stripScaffoldTokens(), a regex-based sanitizer matching only
bracket+pipe token shapes ("<|word|>", "<|word>", "<word|>"), so real
HTML-like text and markdown tables are untouched. Applied to both
text and reasoning segments in getMessageSegments; segments that
become empty after stripping are dropped entirely.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Contributor
|
PR author is not in the allowed authors list. |
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
<|channel|>or the malformed<channel|>, directly in regular text/reasoning output instead of a structured field. Nothing sanitized this before rendering, so raw protocol tokens showed up verbatim in the chat UI.stripScaffoldTokens(), a regex-based sanitizer matching only bracket+pipe token shapes (<|word|>,<|word>,<word|>), so real HTML-like text and markdown tables are untouched.getMessageSegments; segments that become empty after stripping are dropped entirely.Test plan
bun run typecheck,bun run lint,bun run fallow— cleanbun test(app package) — clean, 11 new tests🤖 Generated with Claude Code