CtrlK
BlogDocsLog inGet started
Tessl Logo

tidb-test-diff-triage

Triage unexpected TiDB test diffs that seem unrelated to the current PR. Use when plan/result/testdata changes appear after merge/rebase or only in specific local runs, especially to quickly rule in/out failpoint enablement issues.

87

1.39x
Quality

83%

Does it follow best practices?

Impact

95%

1.39x

Average score across 3 eval scenarios

SecuritybySnyk

Passed

No known issues

SKILL.md
Quality
Evals
Security

Evaluation results

99%

40%

Unexpected Planner Test Diff After Rebase

Failpoint-first triage for planner test diffs after rebase

Criteria
Without context
With context

Failpoint checked first

50%

100%

Build tags do not enable failpoints

8%

100%

References testing-flow.md

0%

100%

Failpoint-enabled rerun with -count=1

12%

100%

Setup issue classification

70%

100%

Symptom section present

100%

100%

Scope section present

100%

100%

Failpoint check section present

62%

87%

Conclusion section present

87%

100%

Action section present

87%

100%

No premature testdata update

100%

100%

90%

15%

Mysterious Executor Test Failures After Weekend Merge

Git bisect workflow to isolate first bad commit in merged range

Criteria
Without context
With context

Minimal reproduction first

100%

100%

git bisect start

100%

100%

git bisect bad with commit

100%

100%

git bisect good with commit

100%

100%

First bad commit identified before testdata update

66%

100%

Symptom section present

42%

100%

Scope section present

57%

100%

Failpoint check section present

28%

85%

First bad commit section present

25%

50%

Conclusion section present

87%

50%

Action section present

75%

87%

No premature testdata sync

100%

100%

96%

25%

Optimizer Plan Change After TiDB Upstream Sync

Testdata update decision after confirmed upstream optimizer change

Criteria
Without context
With context

Failpoint checked first

0%

100%

Root cause proven before sync

100%

100%

Upstream behavior change as valid reason

100%

100%

Semantics unchanged check

100%

100%

Rejects premature update

100%

100%

Symptom section present

42%

100%

Scope section present

33%

100%

Failpoint check section present

0%

85%

Conclusion section present

100%

100%

Conclusion is upstream behavior change

62%

100%

Action section present

80%

60%

Action is sync expected

100%

80%

Repository
pingcap/tidb
Evaluated
Agent
Claude Code
Model
Claude Sonnet 4.6

Table of Contents

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.