用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt;运行前请先检查来源。
npx skills add https://github.com/coreos/ai-helpers --skill pipeline-jira命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地?可下载 SkillsMP 当前能够提供的文件。
正在显示 SKILL.md
OCP release queries - latest versions, RHCOS images, and RPM package lists
Investigate initramfs issues - extraction, module analysis, and comparing working vs failing builds
Deduplication logic for CI pipeline failures - check Jira for existing tracking with semantic analysis
基于 SOC 职业分类
| name | pipeline-jira |
| description | Create JIRA issues for CI pipeline failures - COS project conventions and subtask structure |
JIRA CLI commands and COS project conventions for tracking CI pipeline failures.
Related:
pipeline-failures
# List issues with JQL query
jira issue list --project COS --type Task -q "summary ~ 'Pipeline Monitoring'" --plain
# List open issues
jira issue list --project COS --status "Open" --plain
# List issues by component
jira issue list --project COS --component RHCOS --plain
# Create a sub-task
jira issue create --type Sub-task --parent <PARENT-KEY> --project COS \
--summary "<job> #<build-number> - <stream> <brief-description>" \
--label <label> \
--body "<detailed-markdown-description>" --no-input
# Create a task
jira issue create --type Task --project COS \
--summary "<summary>" \
--body "<description>" --no-input
# Add a comment (comment text is a positional argument, NOT --body flag)
jira issue comment add <ISSUE-KEY> "<comment-text>"
# Multi-line comment
jira issue comment add <ISSUE-KEY> $'Line one\n\nLine two'
# View issue details
jira issue view <ISSUE-KEY>
# Transition issue status
jira issue move <ISSUE-KEY> "In Progress"
⚠️ Important:
jira issue comment addtakes the comment as a positional argument, not a--bodyflag. The--bodyflag is only forjira issue create.
The COS project is used for CoreOS-related issues.
Weekly Pipeline Monitoring tasks track CI failures.
Naming convention: Pipeline monitoring - Sprint NNN - Ws YYYYMMDD
Ws = Week starting (Monday)Find current week's monitoring task:
# Calculate Monday of current week
DOW=$(date +%u)
if [ "$DOW" -eq 1 ]; then
MONDAY=$(date +%Y%m%d)
else
MONDAY=$(date -d "last monday" +%Y%m%d)
fi
# Find this week's monitoring task
PARENT=$(jira issue list --project COS --type Task \
-q "summary ~ 'Pipeline monitoring' AND summary ~ '$MONDAY'" \
--plain --no-headers | head -1 | awk '{print $2}')
Load pipeline-dedup skill for three-pass deduplication logic (exact match, similar failure, semantic analysis) before creating subtasks.
Each build failure should be its own sub-task (including retries that failed).
Summary format:
<job> #<build-number> - <stream> [arch] <brief-description>
Important: Always use the full Jenkins stream (e.g., 4.22-9.8 not 4.22). The stream value comes from the Jenkins build parameters (STREAM or RELEASE). For build-arch, include the architecture after the stream.
Examples:
build #3456 - rhel-9.6 kernel regression in selinux testbuild-arch #1234 - c9s s390x compose failure - repo timeoutbuild-node-image #4216 - 4.20-9.6 TLS handshake timeoutrelease #789 - rhel-9.8 extensions-container build failed## Build Details
- **Job**: <job-name>
- **Build**: #<build-number>
- **Stream**: <stream>
- **Architecture**: <arch>
- **Timestamp**: <timestamp>
- **Duration**: <duration>
- **Jenkins URL**: <url>
## Root Cause Analysis
- **Classification**: <infrastructure | flake | test_regression | package_change | registry_auth | tooling | unknown>
- **Confidence**: <low | medium | high>
<detailed explanation of what caused the failure, >
```
## Labels
| Label | When to Use |
|-------|-------------|
| `flake-infrastructure` | Transient infrastructure issues (repo timeouts, GitHub 500, network) |
| `flake-test` | Flaky test failures (passed on rerun) |
| `bug` | Actual bugs requiring code fixes |
## Issue Linking
For CVE tracking, link OCPBUG issues to RHEL vulnerability issues:
```bash
# Links are created via API, type: "Blocks"
# OCPBUG blocks RHEL (outward)
# RHEL is blocked by OCPBUG (inward)