Testing and code quality

Used here, in this category

• Playwrightchecks
  Fourteen viewports, both sides of every breakpoint. Added after a phone rendered 517 pixels of content inside 375.
  Evidence: scripts/check-mobile.mjs, check-print.mjs, check-visible.mjs
• pa11ychecks
  Runs axe. Caught four keyboard traps the day the figures were repaired.
  Evidence: scripts/check-a11y.mjs
• Lighthousechecks
  Evidence: scripts/check-lighthouse.mjs

All 66

Alphabetical. Checked on 2026-09-04; anything that did not answer says so.

• AICodeSlopMonitor
  Scans public repositories and reports how much of the code looks machine-generated.
  Measures the thing everyone is arguing about.
• Amazon CodeGuru Reviewer
  Amazon's automated review service. The address now serves CodeGuru Profiler, which is about runtime cost rather than review.
  The link has moved on to a different product.
• Applitools
  Visual regression testing. Compares rendered screenshots rather than the DOM, so a layout that moves is caught even when the markup did not change.
  Diffs pixels, because the DOM lies about what a page looks like.
• Autify
  Test automation aimed at quality teams rather than at engineers, with tests authored and maintained without code.
  Written for the QA seat, not the IDE.
• Beeceptor
  Mock API endpoints created in seconds, generated from a prompt or from captured traffic, across REST, GraphQL and gRPC.
  A fake backend before the real one exists.
• BitDive
  Records real Java runtime behaviour and turns it into regression tests and review verdicts.
  Tests written from what production actually did.
• Blackfire.io
  Continuous profiling for PHP and other runtimes, aimed at finding the slow function rather than the slow request.
  Profiles in production, not on a laptop.
• Checksum AI
  Generates, runs and maintains end-to-end tests automatically, and claims to resolve most of its own failures.
  Aimed at the maintenance cost, not the authoring cost.
• Chromatic
  Visual, interaction and accessibility review for component libraries, built by the Storybook maintainers.
  Review at the component, before it reaches a page.
• Codacy
  Automated quality and security review with policy enforcement, aimed at teams that need a gate rather than advice.
  Governance for code nobody read before merging.
• Codeflash
  Searches for a faster version of Python code and verifies the replacement still behaves identically before proposing it.
  Optimisation with a correctness proof attached.
• CodeQL (GitHub)
  GitHub's query language for code. Treats a codebase as a database and a vulnerability class as a query you can write yourself.
  Security findings you can express as a query.
• CodeRabbit
  An automated pull request reviewer that comments line by line and holds a conversation about its own findings.
  Betting that reviewing is the bottleneck, not writing.
• CodSpeed
  Continuous performance testing in CI, measuring instruction counts so results do not move with runner noise.
  Benchmarks that survive a noisy CI machine.
• Coval
  Testing and evaluation for voice agents, covering the full path from simulation to production review.
  Voice agents fail in ways a text test suite cannot see.
• Cypress
  Browser testing that runs inside the page, giving direct access to application state during a test.
  Ran in the browser when everyone else drove it from outside.
• DebuggAI
  Exercises a web application the way a user would and reports what breaks, with no test code written.
  Testing without a test suite.
• DeepSource
  Automated review and static analysis positioned around the volume of machine-written code now reaching pull requests.
  Aimed at the review queue, which is where the new bottleneck is.
• DiffBlue Cover
  Writes Java unit tests by reinforcement learning against the code, verifying each one compiles and passes before proposing it.
  Not a language model, which is the point it makes.
• Ellipsis
  Runs managed agents defined as config files in your repository, reviewing pull requests and building features.
  The agent's configuration is reviewed like any other code.
• Embold
  Static analysis focused on structural technical debt and anti-patterns rather than only on defects.
  Measures the shape of the codebase, not only its bugs.
• ESLint
  The pluggable JavaScript linter. Its rule set became the medium teams argue about style in.
  Where JavaScript conventions are actually written down.
• Fiberplane
  Tooling for operating and observing agent systems, from a team that previously built collaborative incident notebooks.
  Control plane thinking, applied to agents.
• Greptile
  Pull request review with an index of the whole repository behind it, so comments can reference code outside the diff.
  Reviews the diff against the codebase, not against itself.
• Jest
  The JavaScript test runner that bundled assertions, mocking, snapshots and coverage into one install.
  Made testing JavaScript a single dependency.
• JetBrains Qodana
  Takes the inspections from JetBrains IDEs and runs them in CI, so the pipeline enforces what the editor suggests.
  The editor's warnings, made blocking.
• Katalon Studio
  Test management and automation for web, mobile, API and desktop in one platform, aimed at QA organisations.
  Covers desktop, which almost nothing else does.
• KushoAI
  Agents that live in CI and handle testing, self-healing and monitoring rather than being invoked by a person.
  Agents as pipeline residents.
• Launchable
  Predicts which tests are worth running for a given change, cutting execution time by skipping the rest.
  Bets that most of your suite is irrelevant to most changes.
• Lightrun
  Adds logs and metrics to a running application without redeploying, for the bug that only happens in production.
  A print statement you can insert into production.
• Loadmill
  Turns written requirements and tickets into executable test flows for web and mobile.
  Starts at the ticket, not at the selector.
• Mabl
  Low-code test automation with self-healing locators, aimed at keeping a suite running through interface change.
  Assumes the tests will break and plans for it.
• Meticulous
  Generates and maintains front-end browser tests automatically from real usage, without written assertions.
  Coverage you did not have to author.
• MiniMax Code Review
  A GitHub Action that runs code review through MiniMax models.
  Review as a workflow step, not a service.
• Multiplayer
  Connects a coding agent to production signals so it fixes real defects rather than producing speculative changes.
  Aimed squarely at agent-generated noise.
• OctoMind the domain does not resolve
  Generated and maintained Playwright tests with automated test case discovery. The domain no longer resolves.
  Gone: the domain does not resolve.
• Parasoft
  Enterprise test automation spanning static analysis, unit, API and functional testing, common in regulated industries.
  Built for the audit, which is why it is thorough.
• Percy
  Visual regression testing, now part of BrowserStack, comparing rendered snapshots across builds.
  Visual diffing, absorbed into a larger grid.
• Pixee
  Application security that filters findings and opens fixes as pull requests, selling on noise reduction rather than on detection count.
  Aimed at the false-positive problem, not the detection problem.
• Playwright
  Browser automation across Chromium, Firefox and WebKit, with a single API and real waiting instead of sleeps.
  The one that made cross-browser testing boring.
  Used here: scripts/check-mobile.mjs, check-print.mjs, check-visible.mjs
• Prettier
  The opinionated formatter that ended style arguments by refusing to have options.
  Won by giving you nothing to configure.
• PullRequest
  Automated review combined with human expert oversight, so findings are checked before they reach the author.
  Puts a person behind the bot.
• Qodo (CodiumAI)
  Formerly CodiumAI. Generates tests and reviews changes, with the emphasis on covering behaviour rather than producing code.
  Started at the tests and worked backwards.
• Qodo PR Agent
  The original open-source pull request reviewer, still maintained separately from the commercial product built on it.
  The open version that started the commercial one.
• Rainforest QA the site did not answer
  Combined automated testing with crowd-sourced manual testers. The site redirects endlessly and will not load.
  Unreachable: the redirects do not terminate.
• Reflect
  Codeless web and mobile testing, now part of SmartBear, with tests recorded rather than written.
  Absorbed into a larger testing portfolio.
• Relay
  A repository with no published description. The catalogue entry carries little beyond the link.
  Listed, but the project says nothing about itself.
• Rookout
  Live debugging with non-breaking breakpoints, collecting data from running code without stopping it.
  A breakpoint that does not pause production.
• Semgrep
  Static analysis with rules that look like the code they match, so a team can write its own checks in an afternoon.
  Custom rules you can actually read.
• Sherlo
  Visual regression testing for React Native, comparing screenshots on iOS and Android simulators.
  Covers mobile, where most visual tooling stops.
• Snyk (DeepCode)
  Dependency, container and code scanning, sold around fixing rather than reporting, and now around what machine-written code introduces.
  Made the vulnerability database a build step.
• Sonar
  Code quality and security verification, the hosted side of the SonarQube family.
  The quality gate a lot of enterprises standardised on.
• SonarQube
  Self-hosted static analysis with quality gates that can block a merge, now positioned against machine-written code.
  The gate, running on your own server.
• Sourcery
  Automated review in GitHub, GitLab or the editor, suggesting refactors rather than only flagging problems.
  Proposes the rewrite, not just the complaint.
• Sweep
  An assistant for JetBrains IDEs that turns issues into pull requests.
  Aimed at the JetBrains side, which most agents ignore.
• Synth
  A local-first research and experimentation workshop for running many trials on your own machine.
  Research iterations measured per minute.
• SynthScan
  Scans a repository and estimates how much of it was machine-generated.
  Puts a number on a question everyone is asking informally.
• Testim
  Recorded UI tests with machine-assisted locators, aimed at keeping tests alive when the interface shifts underneath them.
  Built around the fact that selectors rot.
• TestRigor
  End-to-end tests written in plain English, with locators inferred so tests survive markup changes.
  Tests that do not name a single CSS selector.
• Testsigma
  Agentic test automation that generates, runs, heals and diagnoses tests, then scores release confidence.
  Reports a confidence number, not just a pass rate.
• Veracode
  Enterprise application security testing with risk management, common where an external attestation is required.
  Bought for the report as much as the scan.
• VibeDoctor
  Scans applications built quickly with assistants for security, performance and quality problems, with uptime monitoring attached.
  Named for exactly the codebase it expects to find.
• Vitest
  A test runner built on Vite, sharing the project's own transform pipeline so tests and source are configured once.
  Reuses the build you already have.
• What The Diff
  Summarises a pull request in plain language, aimed at the reviewer facing a large diff.
  Explains the diff before you read it.
• Wopee.io
  Explores a web application from a URL, generates Playwright tests, runs them and repairs them when the interface changes.
  Starts from a URL and ends with a maintained suite.
• Z.ai Code Review
  A GitHub Action running code review through Z.ai models.
  Review as a workflow step.

Other shelves