AI enhances software testing by automating tasks and generating test ideas, but human expertise remains crucial for strategic analysis and system-level insights. Tools like Trouser bridge human and AI collaboration, offering advanced testing capabilities while safeguarding against misuse.

The EvilTester Show

The Evil Tester Show - hosted by Alan Richardson. Occasional guest episodes. Show notes have comprehensive descriptions and resource links. Released as audio and video.

• https://eviltester.com/show
• https://eviltester.com

Episode Details

• Show Notes: https://eviltester.com/show/032-trowser-rikard-edgren/
• Published: 2026-07-28T09:52:25Z
• Duration: 00:57:17
• Author: Alan richardson

Overview

The discussion centers on the evolving role of AI in software testing and the development of specialized tools to enhance testing practices. Rickard Edgren, a dedicated testing practitioner since 1998, emphasizes that testing is a multidisciplinary activity requiring diverse skills, including technical, domain, and analytical abilities. While AI shows promise in automating unit and integration tests, generating test ideas, and performing detailed security testing - sometimes outperforming humans in vulnerability detection - it is viewed as a complementary tool rather than a replacement for human judgment. AI’s limitations include a lack of strategic depth, over-reliance on token-efficient outputs, and difficulty interpreting system-level behavior from code alone. Testers are encouraged to guide AI with clear requirements and focus on strategic analysis, while leveraging AI for parallel exploration and repetitive tasks.

A key highlight is Trouser, a Chromium-based browser specifically designed for testing, which provides built-in capabilities such as request/response interception, mocking, session sharing, and a REST API for AI integration. Written in C# and currently Windows-only, Trouser enables deeper control than traditional developer-focused tools and supports collaboration between human testers and AI agents. The tool includes safeguards like domain scoping to limit AI actions and reduce security risks. Although not open-source due to concerns about misuse and security, Trouser is freely available and aimed at power users and exploratory testers. The discussion also touches on accessibility testing with tools like axe-core, the challenges of maintaining code quality without extensive unit tests, and the importance of custom, niche tools in advancing testing efficiency and innovation.

What If

• What if you built a custom AI-augmented testing browser for your niche software product?

  • Move: Fork or extend Trouser’s capabilities (e.g., via its REST API) to inject domain-specific test logic for your app - like simulating error states, mocking responses, and auto-running axe-core checks on every navigation.
  • Why Now?: AI agents can now drive browsers via APIs, and you’re already using LLMs (like Opus) for development; combining them with a purpose-built browser lowers integration friction and accelerates test execution.
  • Expected Upside: Cut regression testing time by 50%+ and free yourself to focus on exploratory testing while AI handles repetitive validations in a secure, scoped environment.

• What if you shifted from writing test cases to guiding AI agents in exploratory testing sessions?

  • Move: Use your existing LLM (e.g., Claude Opus) to generate real-time test ideas and payloads during manual sessions in Trouser, then record and refine findings using an LLM-powered note-taker (like the Chrome extension mentioned).
  • Why Now?: AI now outperforms humans in detail-heavy tasks like header manipulation and payload injection, and your solo workflow allows rapid iteration without team alignment overhead.
  • Expected Upside: Discover 2 - 3x more edge-case bugs per session while building a reusable repository of AI-guided test patterns tailored to your software.

• What if you monetized your testing expertise through a micro-tool suite instead of open-sourcing your main project?

  • Move: Package small, high-value plugins (e.g., security header tamper tool, accessibility snapshot reporter) based on Trouser or Playwright + axe, and sell them as standalone tools or subscriptions.
  • Why Now?: You’ve already built 50+ experimental tools with AI help, and niche tools solve real pain points that generic platforms miss - especially for indie devs needing automation without bloat.
  • Expected Upside: Generate recurring revenue while retaining control, avoiding misuse risks of open-sourcing Trouser, and gaining user feedback to inform your core tool’s evolution.

Takeaway

• Download and experiment with the Trouser browser for Windows to streamline testing tasks like request mocking, network interception, and session sharing, especially when combined with Playwright or axe-core.
• Integrate axe-core into your existing test automation stack (e.g., via Playwright or CLI) to automatically catch accessibility issues during development and CI/CD.
• Use powerful LLMs like Claude Opus (or free alternatives like Open Code) to generate and refactor small testing tools or scripts, but validate outputs and prioritize human-guided prompts for better results.
• Build or customize lightweight, niche testing tools (e.g., Chrome extensions for exploratory testing notes) that solve specific workflow gaps instead of relying solely on commercial test management platforms.
• Improve discoverability of your own tools or projects by creating SEO-optimized landing pages - avoid burying them in zip files or deep in documentation - to increase adoption and user feedback.