# Argus — Testing for Coding Agents

> Argus — Testing for Coding Agents is an open-source MCP server for developer tools. MCP server that lets coding agents test your web app in a real browser and report. It runs locally on your machine and is started by your AI client.

- **Kind:** MCP server
- **Pricing:** Open source
- **Platforms:** CLI, macOS, Linux, Windows, self-hosted
- **Website:** https://yichenwu.dev/projects/argus
- **Repository:** https://github.com/chriswu727/argus
- **MCP Registry name:** io.github.chriswu727/argus
- **Categories:** [developer tools](https://ai-powered.com/categories/developer-tools)
- **Listing updated:** 2026-11-18

## About

The README says Argus gives an existing coding agent browser tools and a testing workflow, so the agent decides which journeys to exercise while Argus handles browser actions, repeatable checks and result records. Each suspected failure is rechecked on a fresh page load before it is counted as reproduced, and findings are kept apart from observations and inconclusive checks.

Reports include expected and actual behavior, reproduction steps, screenshots and a record of which testing goals were covered, blocked or left unfinished. Supported findings can be exported as standalone Playwright tests that should fail on the bug and pass once it is fixed. Output is available as HTML, JSON, JUnit and SARIF.

It suits developers who already use Claude Code, Codex or Cursor and want an agent to check their local or staging app. The project describes itself as alpha, and the README advises using test accounts with data that can be changed, since replaying a journey can repeat its writes.

## Features

- **Agent-driven journey exploration:** The connected agent chooses which user flows to test while Argus supplies the browser controls.
- **Fresh-load verification:** Suspected failures are rechecked on a fresh page load before being marked as reproduced.
- **Evidence-backed bug reports:** Findings include expected and observed behavior, steps, screenshots and verification results.
- **Coverage tracking:** Reports show which testing goals were exercised, blocked or left unfinished.
- **Exportable Playwright tests:** Verified findings can be exported as standalone regression tests for a CI suite.
- **Multiple report formats:** Results are output as HTML, JSON, JUnit and SARIF.

## Good for

- Check a local web app's main user flows after a coding session
- Confirm whether a deleted record stays gone after a page reload
- Turn a reproduced UI bug into a Playwright regression test
- Review which testing goals an agent completed and which it left open

## Install

### Claude Code (Terminal)

```sh
claude mcp add argus -- uvx argus-testing
```

### Claude Desktop (claude_desktop_config.json)

```json
{
  "mcpServers": {
    "argus": {
      "command": "uvx",
      "args": [
        "argus-testing"
      ]
    }
  }
}
```

### Cursor (~/.cursor/mcp.json)

```json
{
  "mcpServers": {
    "argus": {
      "command": "uvx",
      "args": [
        "argus-testing"
      ]
    }
  }
}
```

### VS Code (.vscode/mcp.json)

```json
{
  "servers": {
    "argus": {
      "type": "stdio",
      "command": "uvx",
      "args": [
        "argus-testing"
      ]
    }
  }
}
```

### Windsurf (~/.codeium/windsurf/mcp_config.json)

```json
{
  "mcpServers": {
    "argus": {
      "command": "uvx",
      "args": [
        "argus-testing"
      ]
    }
  }
}
```

### Other clients (your client's MCP settings (most use this shape))

```json
{
  "mcpServers": {
    "argus": {
      "command": "uvx",
      "args": [
        "argus-testing"
      ]
    }
  }
}
```

## FAQ

### How do I connect Argus to my coding agent?

The README gives a command for Claude Code: claude mcp add argus -- uvx --from argus-testing argus-mcp. It also gives a Codex equivalent and says the first session downloads Chromium once, about 150 MB.

### Does Argus fix the bugs it finds?

No. The README says a persistence finding confirms the symptom but does not identify the root cause or fix the application.

### Which exported tests are supported?

The README says test exports currently support findings verified through fresh-load text checks. Uploads, restored browser state and other unsupported steps need manual translation.

### Which license does Argus use?

The README shows a license badge linking to a LICENSE file, but the license name is not stated in the content provided.

### Is Argus — Testing for Coding Agents free?

Yes. Argus — Testing for Coding Agents is open source, so you can use and self-host it for free.

### How do I add Argus — Testing for Coding Agents to Claude Code?

Run this command in your terminal: claude mcp add argus -- uvx argus-testing

### What are the best alternatives to Argus — Testing for Coding Agents?

The closest alternatives to Argus — Testing for Coding Agents are Agent Theatre, Momentic, BugMojo, Argentic and agent-config.

## Where this comes from

- Description: written from the project's README, the project's website and its MCP Registry entry.
- Install steps: from the official MCP Registry entry.

## Alternatives

- [Agent Theatre](https://ai-powered.com/mcp/agent-theatre): Agent Theatre is an open-source MCP server for developer tools.
- [Momentic](https://ai-powered.com/mcp/momentic): Momentic is a freemium MCP server for developer tools.
- [BugMojo](https://ai-powered.com/mcp/bugmojo): BugMojo is a freemium MCP server for developer tools.
- [Argentic](https://ai-powered.com/mcp/argentic): Argentic is a freemium MCP server for ai models and llm tooling.
- [agent-config](https://ai-powered.com/mcp/agent-config): agent-config is an open-source MCP server for developer tools.
- [PlaywrightGen](https://ai-powered.com/mcp/playwrightgen): PlaywrightGen is a freemium MCP server for developer tools.
- [Rubric](https://ai-powered.com/mcp/rubric): Rubric is a freemium MCP server for developer tools.
- [Thinkrun](https://ai-powered.com/mcp/thinkrun): Thinkrun is a freemium MCP server for browser automation.

---
Source: https://ai-powered.com/mcp/argus-testing-for-coding-agents
