Skip to content

perf(web): reasoning rows skip the tool-failure text scan #21146

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

Merged
juliusmarminge merged 1 commit into from
Oct 18, 2026
Merged

perf(web): reasoning rows skip the tool-failure text scan #21146

Show file tree
Hide file tree
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
13 changes: 13 additions & 0 deletions packages/client-runtime/src/work-log/presentation.test.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -78,6 +78,19 @@ describe("workEntryIndicatesToolFailure", () => {
},
);

it("ignores failure phrases in reasoning text", () => {
const entry: WorkLogPresentationEntry = {
...base,
label: "Thinking",
tone: "thinking",
itemType: "reasoning",
toolLifecycleStatus: "completed",
detail: "That exit code 2 is just from listing the deleted directory, which is expected.",
};
expect(workEntryIndicatesToolFailure(entry)).toBe(false);
expect(workEntryDisplayIndicatesToolFailure(entry)).toBe(false);
});

it("is true for error tone", () => {
expect(
workEntryIndicatesToolFailure({
Expand Down
5 changes: 4 additions & 1 deletion packages/client-runtime/src/work-log/presentation.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -365,7 +365,10 @@ function workEntryIndicatesToolFailureFromOutput(
) {
return true;
}
if (!workLogEntryIsToolLike(entry)) return false;
// Reasoning detail is the model's own prose (often tens of KB), not tool
// output: scanning it is slow on long threads and misreads phrases like
// "that exit code 2 is expected" as a failed tool.
if (!workLogEntryIsToolLike(entry) || entry.itemType === "reasoning") return false;
if (
resolveT3McpToolDefinition(workEntryToolName(entry)) &&
t3ToolResultIndicatesFailure(workEntryToolOutput(entry))
Expand Down
Loading