Browser tests that heal themselves.
Record tests visually. Self-heal broken selectors. Export to Playwright or Cypress. All running locally on your Mac.
End-to-end tests are the most valuable tests a team can write and the ones they write least. CSS selectors break on every refactor. Enterprise tools put self-healing behind a quote-only sales call. PiperFlow queries Chrome's real accessibility tree, heals broken selectors in 5-15ms with zero external calls, and exports clean code for your CI pipeline. No cloud dependency, no seat-based pricing, no vendor lock-in. PiperFlow is part of ToolPiper Max at $49/month.
Why AX-Native Testing?
Every major testing tool targets the DOM, which is a build artifact. PiperFlow targets the accessibility tree, which represents what users actually see.
Real Accessibility Tree
PiperFlow queries Chrome's native AX tree via CDP's Accessibility.queryAXTree. Not a DOM simulation, not injected JavaScript scanning every element. The actual computed accessibility tree that screen readers consume.
Self-Healing Selectors
Three healing modes: passive quality improvement upgrades weak selectors silently, AX fuzzy matching resolves breaks in 5-15ms locally, and AI-assisted healing handles the rest. Enterprise-grade healing, zero enterprise pricing.
Framework-Proof
A React-to-Vue migration doesn't touch the AX tree if the UI looks and behaves the same. CSS refactors, component library swaps, build tool upgrades, none of it breaks AX selectors. Tests break when behavior changes, which is exactly when they should.
Everything You Need
Full testing toolkit built on raw CDP. No wrappers, no abstractions, no compromises.
Visual Recording
Browse your app normally. Every interaction becomes an AX-enriched test step with element metadata, page context, and a mutation diff showing what changed. No annotation, no code, no switching between panels.
7 Assertion Types
Visible, hidden, text content, URL match, element count, attribute value, and console message. All assertions use polling with configurable timeouts and capture an AX snapshot on failure for debugging.
Temporal Assertions
Three modes for time-dependent verification: always (condition holds for a duration), eventually (condition becomes true within a deadline), and next (condition holds on the very next check). No brittle waits.
Health Monitors
Passive console error, JavaScript exception, and HTTP error checking runs after every test step. Zero configuration. If your app throws a console error during a test, you'll know.
Coverage Reports
Three coverage dimensions in one weighted report: PiperProbe element coverage (60%), CDP JavaScript coverage (30%), and CSS coverage (10%). Color-coded bar with expandable uncovered elements per page.
Playwright / Cypress Export
One-click deterministic export. AX selectors map to each framework's native format: role:button:Sign In becomes page.getByRole('button', { name: 'Sign In' }) in Playwright. Clean, idiomatic code ready for CI.
20 MCP Tools for AI Agents
Any MCP-capable AI client (Claude Code, Cursor, Devin) can create, run, heal, and export tests.
browser_snapshot
Returns Chrome's real AX tree as structured plain text. Auto-connects to Chrome on first call. The AI reads the page the same way a screen reader does: semantic roles, names, and states, not raw HTML.
browser_action
Click, fill, select, hover, scroll, and type using AX selectors. Self-healing resolves broken selectors automatically. Returns a structured AX diff showing exactly what changed on the page after the action.
flow_run
Execute a saved PiperFlow session with optional self-healing, health monitoring, temporal assertions, and coverage collection. Returns pass/fail summary with detailed results per step.
flow_export
Export any PiperFlow session to idiomatic Playwright or Cypress code. The AI can generate tests from a browser snapshot, save them, run them, heal failures, and export the final result, all through MCP tools.
PiperFlow vs The Competition
| PiperFlow | Playwright | Cypress | Testim/mabl | |
|---|---|---|---|---|
| Selector strategy | AX tree native (CDP Accessibility.queryAXTree) | DOM-level ARIA simulation (injected JS) | CSS / jQuery / data attributes | DOM + AI healing layer |
| Self-healing | 3 modes: passive, fuzzy AX, AI-assisted. Local, included in Max | None | cy.prompt() cloud AI (rate-limited) | AI-powered. Paid |
| Test format | JSON (visual UI + MCP + executable) | TypeScript/JavaScript code | TypeScript/JavaScript code | Proprietary visual |
| Test authoring | Visual recorder + MCP + inline edit | Code + codegen + Test Agents (v1.56) | Code + Studio + cy.prompt() | Visual recorder + AI |
| Speed per step | 10-50ms (direct CDP WebSocket) | Fast (bundled browser) | ~23% slower than Playwright | Varies (cloud execution) |
| AI integration | 20 MCP tools. Local, no rate limits | 25 MCP tools (no assertions/healing) | cy.prompt() (cloud, rate-limited) | Built-in (closed) |
| Coverage tracking | PiperProbe elements + JS + CSS, weighted | None built-in | None built-in | Limited |
| Export | Playwright + Cypress code | Native runner | Native runner | Proprietary |
| Browser support | Chrome (CDP) | Chromium, Firefox, WebKit | Chrome, Firefox, Edge | Chrome, Firefox, Edge |
| Multi-tab | Yes (CDP page switching) | Yes | No | Varies |
| Price | $49/mo Max | Free (Apache 2.0) | Free runner / $67-267/mo Cloud | Quote only |
| Vendor lock-in | Export anytime. JSON format | Open source, JS ecosystem | Open runner, paid Cloud features | Enterprise contracts |
How It Works
Download ToolPiper
Signed DMG from modelpiper.com/download. One app, nothing else to install.
Open Chrome
PiperFlow auto-connects via Chrome DevTools Protocol. No extensions, no configuration.
Record
Browse your app normally. Every interaction becomes a test step with AX selectors, element metadata, and mutation diffs.
Run
Self-healing fixes broken selectors automatically. Each step executes in 10-50ms. A 20-step test completes in under a second.
Export
One click to Playwright or Cypress code for your CI pipeline. Clean, idiomatic output ready to commit.
Runs Locally
Your pages, form data, and test content never leave your Mac.
Flat Pricing
Self-healing, recording, assertions, and export are all included in ToolPiper Max at $49/month. No per-seat pricing.
No Lock-in
Export to Playwright or Cypress anytime. Your tests are JSON, not proprietary format.
What PiperFlow costs
PiperFlow ships inside ToolPiper. The app and its local model runner are free; PiperFlow itself is part of the Max plan.
Free
forever
- Native llama.cpp engine, runs any GGUF model
- Unlimited model downloads, multi-model switching
- Local OpenAI-compatible API + embeddings
- MCP host: install any MCP server and run its tools through ToolPiper
- All speech: transcription, text-to-speech, voice cloning, dictation, voice chat
- Talk to your Mac: spoken commands across 26 system domains
- Local chat with tools, on any model you run
- Apple Intelligence on the Neural Engine
- Browser basics: connect to Chrome, snapshot a page, list tabs
- Search Console reports: what your site already ranks for
- Free companion apps (VisionPiper, AudioPiper, MediaPiper)
Pro
- Everything in Free
- Browser automation: act, fill, extract, record, intercept, assert and run Autopilot loops in Chrome
- Files, Git and archives: the agent reads, edits, searches and commits
- Clipboard cache tools: the agent reads and writes your clipboard history
- Mac automation: the agent drives apps, types and clicks, arranges windows and Spaces, launches and stops processes, writes defaults and undoes
- Local RAG over your files
- Web scraping and YouTube transcripts
- Screen and image analysis: screenshot, OCR, color pick, webcam, live stream and capture region controls
- Image composition: import, fill, outpaint, free transform, perspective, gradients and text
- Agent speech and recording: record, transcribe, speak and clone a voice from a tool call
- Cloud AI models in chat, voice and the API (bring your own keys)
- Outreach toolkit: SEO tools (SERP capture with AI Overview detection, keyword discovery, rank check), GitHub activity, content queue
- API discovery toolkit + request replay
- Save captured video streams to disk
Studio
- Everything in Pro
- Image editing and generation: adjust, crop, rotate, flip, transform, cut out, inpaint, remove, segment, generate
- Image upscaling (on-device)
- Image workbenches: layered canvases, masks, adjustment layers
- Storyboards and video: shots, narration, timeline, render
- Video upscaling (ANE-native)
- Image to video, Gaussian splats and 3D worlds
- Screen recording, media extraction and hosting for the clips you make, plus live stream recording with rewind
- Audio waveforms and silence detection
- Pose detection (60fps streaming)
Max
- Everything in Studio
- Code (agentic AI code editor)
- PiperFlow (self-healing browser flows)
- Automations: run any tool, command or chat in a project automatically, once, on an interval or on a cron schedule
- Priority support
Frequently Asked Questions
Does PiperFlow replace Playwright?
For test authoring and iteration, yes. PiperFlow's visual recorder, self-healing, and inline editor handle the create-and-debug loop without code. For CI execution, no. PiperFlow exports to Playwright (or Cypress) code that runs in your existing pipeline. Think of PiperFlow as the authoring tool and Playwright as the CI runtime.
Is PiperFlow free?
No. PiperFlow is part of ToolPiper Max ($49/month): the 6 test tools (flow_save, flow_run, flow_list, flow_get, flow_export, flow_delete), sieve self-healing, visual recording, assertions, health monitoring, temporal assertions, coverage, and Playwright/Cypress export. The browser automation it records against is Pro ($10/month); connecting to Chrome, taking an accessibility snapshot and listing pages stay free with no account.
How does self-healing work without AI?
The default healing mode uses local AX fuzzy matching with zero external calls. When a selector fails, PiperFlow takes a fresh AX tree snapshot and searches for nodes matching the original selector's role and approximate name. Candidates are scored by role match, name edit distance, and tree position. High-confidence matches execute automatically in 5-15ms.
Can AI write tests for me?
Yes, through MCP tools. Any MCP-capable AI client takes a browser snapshot (which returns the AX tree as plain text), reasons about what should be tested, generates PiperFlow steps, saves them, runs them, and reports results. This works with Claude Code, Cursor, Devin, or any MCP client.
Why AX selectors instead of data-testid?
data-testid attributes are stable anchors, but they require developers to add them to every testable element, which means test infrastructure leaks into production code. AX selectors use what's already there: the element's role and accessible name. If your app is accessible, it's testable. If an AX selector doesn't resolve, it often means the element isn't accessible, which is a bug worth knowing about.
Stop maintaining broken selectors.
Download ToolPiper and record your first self-healing test in 60 seconds.