Vigilis — Claude Code MCP server: what it does, how to install
Vigilis checks Playwright, Cypress, and Selenium tests, repairing harmless selector changes while refusing to mask a real regression. Its decisions leave receipts that can be checked offline.
claude mcp add vigilis-mcp -e ANTHROPIC_API_KEY=<ANTHROPIC_API_KEY> -- npx -y vigilis-mcpWhat Vigilis does
Vigilis works with an existing test suite and investigates failures. When a selector has changed, it can update the locator and rerun the test. When the application’s behaviour has genuinely regressed, it leaves the assertion intact and fails the gate. That refusal matters: the server does not make a broken test pass by weakening the check.
Use the MCP server to ask an agent to generate tests, investigate failures, or repair test drift from Claude Code and other supported clients. Vigilis also records an ordered receipt of its actions for offline verification; Treeship can add an independent signature. AI-powered tasks require ANTHROPIC_API_KEY, and the MCP client launches npx -y vigilis-mcp. Receipts for test runs can be created and verified without an API key.
What it helps with
- repair a locator after a selector changes
- stop the test gate on a real regression
- create a verifiable record of a test run
- ask an agent to generate and run a test scenario
Example prompts for the agent
- Investigate the failed login test and update the locator only if the selector changed.
- Find out why the checkout test failed; leave its assertion alone if the app has regressed.
- Generate and run a test scenario for my application.
How to install Vigilis
Platform CLI
claude mcp add vigilis-mcp -e ANTHROPIC_API_KEY=<ANTHROPIC_API_KEY> -- npx -y vigilis-mcpReplace <…> with your own values
Platform CLI
codex mcp add vigilis --env ANTHROPIC_API_KEY=<ANTHROPIC_API_KEY> -- npx -y vigilis-mcpReplace <…> with your own values
Config file
{
"mcpServers": {
"vigilis": {
"command": "npx",
"args": [
"-y",
"vigilis-mcp"
],
"env": {
"ANTHROPIC_API_KEY": "<ANTHROPIC_API_KEY>"
}
}
}
}Replace <…> with your own values
Config file
{
"servers": {
"vigilis": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"vigilis-mcp"
],
"env": {
"ANTHROPIC_API_KEY": "<ANTHROPIC_API_KEY>"
}
}
}
}Replace <…> with your own values
Platform CLI
gemini mcp add -e ANTHROPIC_API_KEY=<ANTHROPIC_API_KEY> vigilis npx -y vigilis-mcpReplace <…> with your own values
Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about Vigilis
How do I install Vigilis in Claude Code?
Run claude mcp add vigilis-mcp -e ANTHROPIC_API_KEY=<ANTHROPIC_API_KEY> -- npx -y vigilis-mcp. Replace the placeholder with your API key.
What does Vigilis do?
It investigates browser test failures, repairs safe locator drift, and refuses to alter a test when the application has a real regression. Its actions can be checked against a receipt.
Does Vigilis need an API key?
AI-powered features require ANTHROPIC_API_KEY. Creating and checking receipts for test runs works without an API key.
