Skip to content

test(evals): Add search agent evals for regex log search - #1405

Open
JoshuaKGoldberg wants to merge 5 commits into
mainfrom
joshgoldberg/logs-1013-add-sentry-mcp-evals-for-regex-log-search
Open

JoshuaKGoldberg wants to merge 5 commits into
mainfrom
joshgoldberg/logs-1013-add-sentry-mcp-evals-for-regex-log-search

Conversation

@JoshuaKGoldberg

@JoshuaKGoldberg JoshuaKGoldberg commented Oct 5, 2026 •

Copy link
Copy Markdown
Member

Adds search_events agent evals for log searches that need key://pattern// regex, plus wildcard searches where regex isn't the better choice. The regex cases are expected to fail until the agent's prompt learns the syntax. I'm adding that in other PRs.

Closes LOGS-1013.

Covers log searches that need key://pattern// regex, plus wildcard searches where regex isn't the better choice. The regex cases fail until the search agent's prompt teaches the syntax.

Refs LOGS-1013
@linear-code

linear-code Bot commented Oct 5, 2026

Copy link
Copy Markdown

LOGS-1013

@JoshuaKGoldberg
JoshuaKGoldberg marked this pull request as ready for review October 5, 2026 15:54
@github-actions github-actions Bot added the risk: low PR risk score: low label Oct 5, 2026
@JoshuaKGoldberg
JoshuaKGoldberg added this pull request to stack #1411 October 5, 2026 15:58
…ex cases

Adds evals for (?i) patterns, UUIDs within the 64-character limit,
prefix and either-or searches that should stay wildcards, and spans or
errors searches that must not get a regex.

Refs LOGS-1013
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Regex matchers now reject over-long, lookaround, backreference, and
over-escaped patterns, the non-logs cases must keep the user's text, and
the plain-text guards no longer copy prompt examples.

Refs LOGS-1013
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The regex check now also rejects possessive quantifiers, atomic groups,
and escapes RE2 refuses, and the ignoring-case case accepts [0-9] as
well as \d.

Refs LOGS-1013
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
… file

The possessive check no longer trips on character classes or \p{...}+,
more escapes RE2 refuses are rejected, and the file is biome-formatted.

Refs LOGS-1013
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

This branch was successfully deployed

No deployments
Actions — 17fc7f3e Deployed Oct 5, 2026 by JoshuaKGoldberg via eval #1219
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

risk: low PR risk score: low

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant