microshift-ci:doctor

Analyze MicroShift CI failures across releases and rebase PRs into an HTML report.

7|15|Updated Jan 23, 2026

Search Tags:#root cause analysis #artifact analysis #prow #html reporting #microshift #ci diagnosis #jira correlation

Ready to run

npx skills add https://github.com/openshift-eng/edge-tooling --skill microshift-ci-doctor

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: microshift-ci:doctor
Source: https://github.com/openshift-eng/edge-tooling/tree/main/plugins/microshift-ci/skills/doctor
Command: npx skills add https://github.com/openshift-eng/edge-tooling --skill microshift-ci-doctor

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires gsutil, gh, pcp-export-pcp2json, matplotlib, and includes scripts (resource) components.

What problem does it solve?

When MicroShift CI runs fail across multiple releases (and related rebase PRs), the work of correlating failing jobs, artifacts, root causes, and existing Jira bugs becomes slow and fragmented.

Core Features & Use Cases

Multi-release CI triage with consolidated reporting: Accepts a comma-separated list of MicroShift releases and generates one HTML summary covering both periodic jobs and relevant rebase PR failures.
Deterministic artifact collection and aggregation: Uses scripts to fetch failures, download artifacts, generate graphs (when PCP data is present), and aggregate results into structured outputs.
Per-job root cause analysis and Jira correlation: Runs LLM agents for job-specific investigation and performs bug search (dry-run) so the report includes matching/open bug information.

Quick Start

Run the doctor for releases 4.21 and 4.22 by asking: “/microshift-ci:doctor 4.21,4.22”.

Frequently Asked Questions about microshift-ci:doctor

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

FAQPage Schema

How do I diagnose MicroShift CI failures across multiple releases at once?▼

To diagnose MicroShift CI failures across multiple releases, you can provide a comma-separated list of release versions to consolidate failing jobs, artifacts, root causes, and Jira bug correlations into an HTML report.

Can I correlate existing Jira bugs with failing Prow jobs during CI triage?▼

Yes, you can correlate existing Jira bugs with failing Prow jobs by running a bug search dry-run during the per-job agent-driven root cause analysis, which embeds matching open bug information directly into the final HTML report.

Do I need gsutil and gh installed to collect CI artifacts for root cause analysis?▼

Yes, you need gsutil and gh installed because deterministic scripts use them to fetch Prow failures, download required CI artifact directories, and aggregate data before root cause analysis begins.

What is the best way to generate graphs from PCP data collected during CI failure investigations?▼

The best way to generate graphs from PCP data is using the deterministic prepare, graph, and finalize scripts, which leverage pcp-export-pcp2json and matplotlib to produce visualizations when PCP data is present in the downloaded artifacts.

Does MicroShift CI diagnosis work for both periodic job failures and rebase PR triage?▼

Yes, MicroShift CI diagnosis works for both periodic job failure investigations and PR rebase failure triage, accepting release lists to generate one consolidated HTML summary covering both relevant job types.

Why are my CI artifact directories not being aggregated into the final HTML report?▼

If CI artifact directories are not aggregated into the final HTML report, it is likely because the deterministic scripts could not locate pre-downloaded artifacts in the computed work directory required for the agent-driven analysis.