Piwi Dashboard

🗄️

History that outlives CI

Every run, trace, and HTML report kept and browsable long after the build agent deleted its artifacts — with pass-rate, duration, and stability trends computed across all of them.

🧩

Failure clustering

An error fingerprint collapses forty red tests into the three root causes behind them, grouped across specs and across runs, so one cause is one thing to triage.

📉

Flaky tests, scored and costed

A composite flakiness score with a root-cause class (timing, network, assertion…) and the CI minutes each flake wastes — so you fix the expensive ones, not the annoying ones.

🩹

Locator healing

When a selector breaks, ranked replacements captured from the last passing run, with a recommended fix that matches your existing conventions.

🔬

Evidence in one place

The bundled Playwright trace viewer, screenshots, console, network calls, Web Vitals, and the failing call stack with real source — all served by your own instance.

🔒

Yours to run

One Docker container, SQLite or PostgreSQL, local or S3 storage, optional role-based auth. Zero telemetry — the only outbound calls are the ones you configure.

Why this exists

Playwright's HTML report is excellent, and it lasts exactly until the next build. So the questions that actually matter get hard to answer: Has this test always been flaky? Did my fix work? Which of these forty red tests are the same bug? What did we change the day the suite started failing?

Piwi keeps the runs, so you can answer them — and then does the things a permanent history makes possible. It is deliberately Playwright-only: that is what makes traces, step timing and locator healing first-class rather than lowest-common-denominator. If you need one place for JUnit, pytest and Cypress results too, or you only ever debug locally and never look back, Why Piwi? says so plainly.

See it in action

Live streaming — a run updating in real time as tests complete. No polling, no waiting on CI to finish.

Dashboard overview — portfolio stats, per-project health and recent activity

Dashboard overview — at-a-glance stats, each project's recent runs and pass rate, and live activity across all projects

Project detail — per-run result bars above the run history table

Project detail — one bar per run over the complete history, filtered by environment

Test run detail — every test case with status, duration, location, and error messages

Failure clusters — tests sharing the same root cause are grouped by error fingerprint

Flaky tests — composite flakiness score with retry-pass and alternation detection

Performance — per-run total, average and P90 duration, above the slowest-tests ranking

Cluster triage — set status, write triage notes, and track resolution

Where to go next

Start from what you came here to do.

Something is failing right now — Recipes answer the question you arrived with: is this a regression or a flake, which locator to use instead, forty tests are red, the suite is unreliable, CI takes too long.

Setting it up — Getting started goes from a Docker command to your first run in the dashboard. On a laptop, the desktop app bundles the whole server in a native window: no Docker, no Node.

Learning the vocabulary — Core concepts defines runs, test cases, executions and clusters. Worth five minutes before the rest.

Wiring up CI — CI & sharding: two environment variables, why ten shards are one run, and how to block a merge on the analysis rather than the exit code.

Running it for a team — Deployment, Configuration, Authentication, and Privacy & data flow.

Letting an agent do the reading — the MCP server gives a coding agent 40 tools over your test history, and AI diagnosis explains a cluster against your actual git diff with a provider you configure. Both optional; a local model works.

Piwi DashboardYour Playwright results, kept and explained