report-flaky-test

Report flaky RediSearch CI test failures to Jira with collected logs and triage.

6.2k|596|Updated May 5, 2016
One-click install
npx skills add https://github.com/RediSearch/RediSearch --skill report-flaky-test
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: report-flaky-test
Source: https://github.com/RediSearch/RediSearch/tree/main/.skills/report-flaky-test
Command: npx skills add https://github.com/RediSearch/RediSearch --skill report-flaky-test

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Flaky tests in RediSearch CI waste engineering time and often go untracked or get duplicated in Jira. This Skill standardizes the reporting workflow: it collects failure evidence from GitHub Actions, finds or creates the right MOD Jira ticket, and proposes immediate triage so flaky tests are documented consistently.

Core Features & Use Cases

  • Evidence Collection: Gathers failure details from GitHub Actions runs, job logs, and downloaded test artifacts using the gh CLI, with fallbacks when --log-failed returns empty output.
  • Jira Deduplication: Searches existing MOD tickets by normalized test id and flaky_test label before creating a new Bug under the MOD-9672 epic, always asking for confirmation before mutating Jira.
  • Triage & Skip Workflow: Suggests scoped test-only fixes or temporary skip_until markers, and can prompt to dispatch the flaky-test mark workflow with explicit branch scope approval.
  • Use Case: A CI run fails on test_expire:testExpire only on macOS. Provide the run URL, and the Skill downloads the logs, identifies the likely owner from git history, finds the existing MOD ticket, and prepares a comment with the new occurrence details.

Quick Start

Report the flaky test from this failed GitHub Actions run URL and create or update the matching MOD Jira ticket.

Frequently Asked Questions about report-flaky-test

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I report a flaky test from a failed GitHub Actions run?

Provide the run URL, job URL, or run id, and the Skill uses the gh CLI to fetch job logs and download test artifacts. It then extracts the test id, failure reason, platform, and branch details to build a Jira report.

How to avoid creating duplicate Jira tickets for flaky tests?

The Skill searches Jira with JQL matching the normalized test id and the flaky_test label before creating anything. If an existing ticket matches, it prepares a comment with the new occurrence instead of a new issue.

What should I do when gh run view --log-failed returns empty output?

This happens on some merge_group runs and macOS matrix jobs. Fall back to the REST API with gh api on the job logs endpoint, then grep the downloaded log for failure markers like FAIL or Traceback.

Can the Skill create or update Jira tickets automatically?

No. It always shows the exact issue payload or comment body and asks for confirmation before creating or commenting in Jira. Workflow dispatches for marking tests as flaky also require explicit approval.

When should a flaky test be temporarily skipped instead of fixed?

A temporary skip_until is suggested when the failure repeatedly disrupts CI or the merge queue and no trivial test-only fix is visible. The skip must be scoped to the smallest affected test, mode, or platform with an expiry date and Jira key.