About | Blog | RSS | Medium | Dev.to | WeChat | Zhihu
I build open-source tools for DevOps, CI/CD, and software delivery — collected as the Open Delivery Stack: standards, quality gates, CI feedback, dependency safety, and maintainer automation. AI is increasingly part of this: explaining failures, reviewing changes, and helping teams ship with evidence.
| Area | Problem | Projects |
|---|---|---|
| Define engineering standards | Teams need consistent branch names, commit metadata, and contribution signals before work enters CI. | conventional-branch/conventional-branch, commit-check/commit-check |
| Shift quality left | Developers should catch C/C++, Jenkinsfile, and Dockerfile issues before review. | cpp-linter/cpp-linter-action, cpp-linter/cpp-linter-hooks, jenkinsfilelint, hadolint-pre-commit |
| Secure and sustain dependencies | Supply-chain and runtime-version risks should be visible before they become production problems. | pipguard, py-eol |
| Understand engineering systems | Teams need readable signals about repository health, delivery maturity, and generated quality reports. | gitstats, devops-maturity/devops-maturity |
| Diagnose CI/CD failures with AI | Build failures should come with context, likely causes, and next actions instead of raw logs only. | jenkinsci/explain-error-plugin |
| Automate OSS maintenance | Maintainers need agents that can triage, reason, patch, and keep repositories moving. | repokeeper, aion |
| Standardize team workflows | Platform work should turn one engineer's operational standards into repeatable team pipelines; cast-cli is now in maintenance mode as focus shifts to ODS. | castops/cast-cli, atlassian-api-py |
| Govern delivery with machine evidence | AI writes code faster than governance can track; teams need machine-readable evidence at every stage. | open-delivery-spec/spec, open-delivery-spec/cli, open-delivery-spec/validate-action |
- Open Delivery Spec (ODS) — machine-readable delivery governance: spec, CLI, and CI validation so every commit-to-deploy stage leaves auditable evidence.
- commit-check — configurable Git commit and branch name validation for teams and CI pipelines.
- cpp-linter — C/C++ linting, formatting, and static analysis feedback directly in PRs and pre-commit.
- Explain Error Plugin — AI-powered Jenkins failure diagnosis: causes, explanations, and suggested fixes instead of raw logs.
- repokeeper — agent-based maintainer tooling for triaging issues, reasoning about code, and keeping repositories healthy.
- Maintainer and contributor in the Jenkins ecosystem, including AI-powered failure diagnosis.
- Contributor to PyPA (merged PRs) and Python (merged PRs).
- Maintainer of mkdocs-ng/mkdocs, mkdocs-ng/mkdocs-material, gitstats, and badgepy.
WeChat: shenxianpeng - sharing AI + DevOps practices and real-world engineering notes.
- cpp-linter-hooks: The Most Complete pre-commit Solution for C/C++ Projects
- Conventional Branch Official Skill Is Here—Install with One Command
- mkdocs-ng Maintenance Progress — v1.7.x Fix Summary and Next Steps
- Building an AI Agent Inside Jira —— A Jira Copilot Implementation Guide
- Counter-intuitive Designs in the pi project — From AGENTS.md to "Just Close Your PR First"
- Explain Error Plugin Major Update — AI Auto-Fix, Usage Management, and More AI Providers
- Writing an Article for Twenty-Four Cents—My Experience Using Pi + DeepSeek as a Codex Backup Solution
- AI Agent, or Automation?
- Python Supply Chain 'Nuclear Bomb' Attack Just Happened—A
pip installSteals All Your Credentials! - A Three-Year Community Pain Point—I Put Hadolint on PyPI, Ending Manual Installation