segentic-lab/periscope-mcp — explained in plain English
Analysis updated 2026-05-18
Have an AI agent automatically fill out and test forms on a web app, including sites that need login.
Crawl an entire website and get a report on accessibility, SEO, broken links, and performance issues.
Test authenticated flows, including two factor authentication and single sign on, using a one time interactive login.
Catch silent failures in browser automation, like a drag action that appears to succeed but changes nothing.
| segentic-lab/periscope-mcp | alx-code/lingbot-video-1.3b-fp8 | andrewbergman/kuma-importer | |
|---|---|---|---|
| Stars | 7 | 7 | 7 |
| Language | Python | Python | Python |
| Setup difficulty | moderate | — | moderate |
| Complexity | 3/5 | — | 2/5 |
| Audience | developer | developer | ops devops |
Figures from each repo's GitHub metadata at analysis time.
Requires Python 3.11+ and Playwright with a Chromium browser, an install script handles most of the setup.
Periscope is a tool that lets AI agents test and audit websites and web apps, using a standard called MCP that lets AI assistants like Claude call outside tools. Instead of giving an agent raw browser control and asking it to squint at screenshots, Periscope offers 74 ready made tools that check things and return a clear pass or fail answer, along with the actual value that was found.this includes checking accessibility, SEO, and how well a site can be found and understood by AI powered search tools. The tools are built around the kinds of tasks a real tester would do. One tool can detect and fill out an entire form on its own, guessing sensible values for each field. Another can run a batch of up to 25 actions in one go, like clicking, typing, and checking results, instead of one call per step. A separate tool can crawl a whole website and audit every page for accessibility, SEO, broken links, and performance, saving a report at the end. It can also log into websites that require a username and password, or even handle two factor authentication and single sign on by opening a visible browser window for a human like login step, then continuing headlessly afterward. Under the hood, Periscope runs on Playwright, a browser automation library, controlling a headless version of Chrome. It keeps track of ongoing browser sessions, stores project configurations as JSON files, and saves screenshots, videos, and reports so an agent can review what happened. When something silently fails, such as a drag and drop that does not actually move anything, Periscope flags it as a failure instead of reporting false success, and explains what likely went wrong. It is written in Python, requires Python 3.11 or newer, and depends on Playwright with a Chromium browser. Installation is done through a shell script that sets up a virtual environment, installs dependencies, and configures the tool for use with any MCP compatible AI assistant, such as Claude Code, Cursor, or Windsurf. The full README is longer than what was shown.
An MCP server that gives AI agents 74 tools to test, audit, and QA websites and web apps, returning clear pass or fail results instead of screenshots.
Mainly Python. The stack also includes Python, Playwright, Chromium.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Don't trust strangers blindly. Verify against the repo.