Skip to content
Agents tracked: 492 Releases (7d): 545 Agent status: 1 with issuesSince last limit reset: Claude 9d · Codex 7d For AI agents: Markdown Updated Nov 20, 2026

AI agents for testing

Agents that write unit tests, run your suite, fix what fails and click through your app like a QA engineer.

By weekly package downloads, the leaders are Claude Code (18.9M), Stagehand (2.9M). By VS Code extension installs, the leaders are GitHub Copilot (79.4M), Claude Code (30.6M). By GitHub stars, the leaders are Claude Code (156K), Aider (50.2K), Stagehand (26.1K). Among agents with at least 20,000 downloads in the last 30 days, the fastest grower is Stagehand, with downloads up 40%. As of Nov 20, 2026.

Works with:
Usage
  1. 1
    Anthropic · Terminal · terminal agent + IDE extensions

    Anthropic's agentic coding tool that reads a codebase, edits files and runs commands from the terminal, with VS Code and JetBrains integrations plus desktop, web and mobile access. It runs Claude models through a Claude subscription, an Anthropic Console API key or supported third-party cloud providers.

    CLIIDEDesktop appMCPOwn API keyPlan limits
    88 18.9M up 15.5% 156K yesterday 18.9M downloads a week up 15.5% 30dGid 88 $20/mo
  2. 5
    GitHub · IDEs · AI coding assistant (IDE agent mode, CLI and cloud agent)

    AI coding assistant from GitHub offering completions, chat and agent mode in IDEs, a terminal CLI, a desktop app and a cloud agent that works on assigned tasks in the background and opens pull requests. Models from several providers can be selected.

    IDECLIWeb appMCPLocal modelsOwn API keyPlan limits
    75 — — — — 79.4M VS Code installsGid 75 Free + $10/mo
  3. 21
    Browserbase · Browser · browser automation library

    Open-source SDK from Browserbase for building browser agents that mix natural-language actions (act, extract, observe, agent) with code, available in TypeScript, Python and Go and runnable locally or on Browserbase cloud browsers.

    Open sourceAPI / SDKOwn API key
    57 2.9M up 39.5% 26.1K 2d ago 2.9M downloads a week up 39.5% 30dGid 57 Free + $20/mo
  4. 24
    Cognition · Cloud · cloud coding agent

    Autonomous AI software engineer from Cognition that runs in cloud VMs, takes work from Slack, Teams, Linear, Jira or an API, and delivers pull requests. Devin Desktop and the Devin CLI act as local companions to the cloud agent; the CLI can also create, steer and resume Devin Cloud sessions.

    Web appCLIDesktop appMCP
    56 — — — — 1,322 Homebrew installs a monthGid 56 $20/mo
  5. 45
    Aider-AI · Terminal · terminal agent

    Open-source AI pair-programming tool for the terminal that edits files in a local git repository and commits the changes. It works with multiple LLM providers using the user's own API keys.

    Open sourceCLILocal modelsOwn API key
    50 116K up 33.4% 50.2K 15mo ago 116K downloads a week up 33.4% 30dGid 50 Free (OSS)
  6. 157
    Amazon · Browser · browser automation service

    AWS service and Apache-2.0 Python SDK for building and running fleets of UI-automation agents that complete browser workflows from natural-language act() calls plus Python code and hand off to a human when needed. Workflows are developed in a web playground, IDE extension or CLI and deployed on AWS.

    Open sourceAPI / SDKCLIWeb appMCP
    30 116K up 24.8% 920 4d ago 116K downloads a week up 24.8% 30dGid 30 Usage-based
  7. —
    Anthropic · Browser · browser extension agent

    Anthropic's Chrome extension that lets Claude read pages and click, type, navigate and fill forms in the user's own signed-in browser, with recorded workflows, saved shortcuts and scheduled tasks. In automatic-approval mode a safety check reviews each action and sensitive actions wait for the user; Team and Enterprise admins can set site allow and block lists. It connects to Claude Code, Cowork and the Claude desktop app. Other browsers and mobile are not supported.

    Browser extension
    — — — — — No public usage numbers Free + $20/mo

“#” is the position among products of the same kind (agents, frameworks, tooling) by Gid Score, our 0–100 composite of adoption, community, attention and momentum. Use the filters and sorting above, or switch to the table view. Methodology · All AI agents.

Two kinds of testing agent

Coding agents write and run tests in your repository: they read the code, add unit and integration tests, run the suite and iterate on failures. Browser agents test the running application from the outside, following a user flow described in plain language. The first raises coverage; the second catches what only shows up in a real browser.

What each agent does for this task

Only agents whose own documentation describes this use are listed. Checked Oct 6, 2026.

Agent What it does Source
GitHub Copilot Generates unit and integration tests, including edge cases and exception handling docs
Claude Code With Chrome connected, tests local web apps: form validation, visual regressions, user flows and console errors; Chrome integration is not available through Amazon Bedrock, Google Cloud or Microsoft Foundry docs
Aider Runs your test command with /test or --auto-test and tries to fix the failures docs
Devin Analyses a test suite, writes new tests to raise coverage and verifies them by running the suite docs
Claude in Chrome Browser extension that Claude Code drives to test web apps and check user flows in a live browser docs
Amazon Nova Act UI tests defined in natural language, with visual logs of each run docs
Stagehand Runs end-to-end browser tests with Stagehand or Playwright in isolated cloud browsers docs

Keep the tests honest

  • Watch for tests that test nothing. An agent asked to make tests pass can weaken assertions or mock away the code under test. Review the assertions, not just the green tick.
  • Pin the behaviour first. Ask the agent to write a failing test that reproduces a bug before it fixes it.
  • Browser tests are slower and flakier than unit tests; keep them for the flows that matter most.

More options by job: the best AI agents in every category.