返回资讯中心
外部精选
软件工程
#开发者实时流

Show HN: StackVitals – CI failure triage and DORA metrics, no server

The vital signs of your GitHub repo. Flaky tests, slow reviews, delivery and DORA, risky

Hacker News 最新kaushik-yadav2026/11/1 · 17 天前15 分钟阅读

以下正文同步自 Hacker News 最新,版权归原站所有,已转换为易读排版。

The vital signs of your GitHub repo. Flaky tests, slow reviews, delivery and DORA, risky dependencies and community health, measured in your own GitHub Actions.

A tour on demo data. Watch it with sound.

A free, open-source developer suite for GitHub: CI health and flaky tests, pull request flow, delivery and DORA, dependencies, community health and an org portal. It runs in your own GitHub Actions and keeps its history in your own repo: no StackVitals-hosted backend, account or subscription, and optional integrations use services and keys you choose. GitHub's built-in token covers every module in a repo; an org's private repos in the Portal need a read-only App, and org-wide Copilot usage an org token. Actions minutes on private repos are billed by GitHub as usual (docs/scaling.md). The rules behind its verdicts are checked on public data, with the losses published.

Every run publishes an app on your repo's GitHub Pages (or stackvitals ui owner/repo on your machine): one page per module, decisions and settings saved straight from the browser, setup for more repos in a few clicks once GitHub is connected (connecting takes one token, made from a prefilled GitHub page; Pages are turned on for public repos), and a demo at ?demo=1. docs/app.md walks through it.

Install it as the GitHub Action stackvitals/stackvitals@v0 (also on the GitHub Marketplace), or the CLI and MCP server from PyPI: pip install stackvitals (or run it without installing: uvx stackvitals).

When CI fails on a pull request, one comment sorts each failure, from the repo's own CI history:

1 of 3 failure(s) look like flakes; the rest need a look.

The verdict check still blocks until CI is green: a known flake is re-run, never waved through.

pip install stackvitals gh auth login # once, so StackVitals can use your GitHub sign-in stackvitals init your-name/your-repo

init adds one workflow file to your repo, runs it once and turns on your dashboard:

Public repo: the dashboard goes live at https://your-name.github.io/your-repo/.

Private repo: your data stays in your private repo, and the dashboard opens on your machine:

stackvitals ui your-name/your-repo

Just looking? stackvitals ci report owner/repo reports CI health for any public repo in about 20 seconds, with nothing added to it. Prefer clicking? The app's Setup page does the same as init. More: docs/quickstart.md.

Every step, from what starts a run to what you see: the architecture in detail.

Module What it answers Status

CI Health Which jobs and tests are flaky or slowing down, who owns them, coverage and patch coverage stable

Team Flow Who has to act next on each pull request, and how long reviews take stable

Delivery Cycle time split into coding, pickup, review and deploy (LinearB's phases), DORA's five metrics, AI-assisted work compared like with like stable

Insights What to fix first across modules, ranked by hours lost, with alerts and a weekly digest stable

Dependency Steward Known vulnerabilities and malware in the lockfiles OSV-Scanner reads (npm, PyPI, Go, Rust, Maven, Ruby, .NET, PHP and more; 65 of 100 sampled repos had one); whether to add, keep or upgrade a package, with upgrade evidence from public repos preview

Community Where newcomers drop out, how fast they get an answer, who carries the work stable

Portal A lightweight catalog on GitHub alone: what an org owns, who owns it (checked), scorecards, golden paths and an org-wide view preview

Explore SQL over your repo's data in the browser, with links to share a query preview

Contribute Open-source issues an outsider can realistically get merged stable

Stable: measured, reviewed and used on the sandbox for every release. Preview: built and tested, but newer or measured on less; its behaviour may still change.

Works with Jira Cloud, Linear, PagerDuty, incident.io, Copilot usage reports, Slack, Teams, Discord, Discourse, test reports from other CI systems, Backstage, and DevLake's tables (docs/connectors.md). AI agents get every module through one MCP server (11 core tools by default, all 35 with --tools all).

Everything works with GitHub's built-in token. Integrations are optional; pass their keys as repository secrets:

  • uses: stackvitals/stackvitals@v0 with: webhook: ${{ secrets.SLACK_WEBHOOK }} # alerts to Slack, Teams or Discord pagerduty-token: ${{ secrets.PAGERDUTY_TOKEN }} # incidents count in DORA jira-email: ${{ secrets.JIRA_EMAIL }} # Jira, with its site URL under jira-token: ${{ secrets.JIRA_TOKEN }} # [connectors.jira] in .github/stackvitals.toml

Every input: docs/install.md. Every setting: docs/settings.md.

Your coding agent can ask StackVitals before and while it changes code: is this failure a known flake?, what is blocking PR #123?, who should review this?, is this package safe to add? It answers from your repo's history, with the evidence.

claude mcp add stackvitals -- uvx stackvitals serve

Claude Desktop, Cursor, VS Code and other MCP clients:

{ "mcpServers": { "stackvitals": { "command": "uvx", "args": ["stackvitals", "serve"] } } }

It uses your gh login or GITHUB_TOKEN. The 11 core tools load by default (ci_failure, is_known_flake, pr_queue, who_should_review, check_package, insights and more); stackvitals serve --tools all loads all 35. Listed in the MCP registry as io.github.stackvitals/stackvitals.

Measured on public repos beside the rules other tools use (docs/accuracy.md):

It publishes what failed too. Of the failures an old rule called safe to re-run, 44% (15 of 34) were caused by the change. The tightened rule still got 13 of 81 held-out calls wrong (16%). A stricter one (a flake's sign in the log, no changed file named) got none of 4 fresh calls wrong, too few to show 5%, so nothing is called safe: likely flakes are re-run, never passed by the check. Each sample was labelled by two model families (Claude and Gemini); no person has audited the labels.

What each one does better, what StackVitals does differently, and what was measured side by side (docs/compare):

正文由 探How 从来源站点 RSS 同步,内容未经改写;遇到排版缺失或需要图片、视频时请以原文为准。

打开原文
去社区讨论 →