# Connect Bug Reports to Your AI Tool with clip.qa MCP

To connect bug reports to your AI tool, add clip.qa's remote MCP server to your client or run its local npm package with an API key. clip.qa documents setup for Claude Code, Cursor and Windsurf.

## Remote MCP setup

The remote server uses Streamable HTTP at https://mcp.clip.qa/mcp. Sign in through your browser on first use with OAuth 2.1 and PKCE. It exposes 15 tools.

### Claude Code

Run:

    claude mcp add clip-qa-team --transport http https://mcp.clip.qa/mcp

Claude Code's [MCP documentation](https://code.claude.com/docs/en/mcp) explains its MCP configuration.

### Cursor

Add the server URL to your mcp.json configuration:

    {
      "mcpServers": {
        "clip-qa": {
          "url": "https://mcp.clip.qa/mcp"
        }
      }
    }

Cursor's [MCP documentation](https://cursor.com/docs/context/mcp) covers adding MCP servers.

### Windsurf

clip.qa provides Windsurf steps on its [setup page](https://clip.qa/setup). Follow that guide for the configuration clip.qa supports.

### Remote tools

The remote server includes:

    listReports
    getDetails
    getNetworkRequests
    getConsoleLogs
    getScreenshots
    getVideoTranscript
    analyzeVideo
    searchReports
    getReportDiff
    getProjectSummary
    getFrameAnalysis
    getReportRaw
    createComment
    linkFix
    linkFixAttempt

createComment, linkFix and linkFixAttempt require write scope. getProjectSummary requires project read scope.

## Local MCP setup

Install the local package:

    npm install @clip-qa/mcp-server

Configure it with a clip.qa API key. The local server exposes 11 tools:

    get_report
    get_report_for_llm
    get_report_dom
    get_report_raw
    list_reports
    search_reports
    get_report_diff
    get_project_summary
    create_comment
    link_fix
    link_fix_attempt

The local server provides access to report content, DOM data, raw report data, project summaries and linked fixes. It also supports comments and fix linking when your API key has the required permissions.

## What clip.qa gives an MCP agent

Reports remain private until explicitly shared. PII is scrubbed from MCP output.

An agent can read the report, screenshots, video transcript, frame analysis and report diffs. For Chrome-extension reports, an agent can also read console logs and network requests. clip.qa also provides AI bug reports that combine video and other report data for review.

Video analysis runs in the cloud. There is no SDK to install, and MCP access is included on every plan.

## clip.qa compared with Jam and TestMu AI

| | clip.qa | Jam | TestMu AI |
|---|---|---|---|
| MCP type | Remote: Streamable HTTP. Local: npm package | Streamable HTTP | Streamable HTTP |
| Tools count | 15 remote; 11 local | About 35 | Not stated on page |
| Auth | Remote: OAuth 2.1 with PKCE and browser sign-in. Local: API key | Browser sign-in on first use | OAuth |
| Bug-report focus | Reports, screenshots, video and frame analysis; console logs and network requests for Chrome-extension reports | Chrome extension, Mac and iOS app recordings with console and network logs, recording links and AI summaries | Test runs, automation logs and test manager workflows; bug reports are not mentioned on the MCP page |

[Jam](https://jam.dev/docs/jam-mcp.md) is strong at browser session capture with console and network logs, and its MCP server has the widest tool set in this table.

[TestMu AI](https://www.testmuai.com/mcp/) is focused on running, debugging and triaging test runs across HyperExecute, automation logs and test manager workflows. Its MCP page does not mention bug reports.

[BrowserStack Bug Capture](https://www.browserstack.com/bug-capture) focuses on browser-extension screen recording. [Luciq](https://luciq.ai/pricing) is another option to evaluate, though its pricing page does not publish prices.

## Which clip.qa setup should you use?

Use the remote server when you want OAuth browser sign-in and the full 15-tool set. It works over Streamable HTTP without installing a package.

Use the local npm package when you prefer a local MCP process and API-key authentication. It provides 11 tools for reports, project summaries, comments and fix linking.

For setup steps for Claude Code, Cursor and Windsurf, see the [clip.qa setup guide](https://clip.qa/setup).

## Frequently asked questions

### How do I connect my bug reports to my AI tool MCP?

Add https://mcp.clip.qa/mcp as a remote MCP server in your supported client, or run @clip-qa/mcp-server locally with an API key. clip.qa documents remote setup for Claude Code, Cursor and Windsurf.

### Does clip.qa require installing an SDK?

No SDK is published. You can use the remote MCP server over Streamable HTTP or install the local MCP npm package.

### What bug-report data can an AI agent read?

An agent can read reports, screenshots, video transcripts, frame analysis and report diffs through the remote MCP tools. For Chrome-extension reports, it can also read console logs and network requests. Local setup provides report, DOM, raw report, project summary and fix-linking tools.

### Can clip.qa write back to a report?

Yes. The remote createComment, linkFix and linkFixAttempt tools require write scope. The local package also includes create_comment, link_fix and link_fix_attempt.

### Is MCP included in every clip.qa plan?

Yes. MCP is included on every plan: Free, Team and Enterprise.

## Sources

- [clip.qa setup page](https://clip.qa/setup)
- [Claude Code MCP documentation](https://code.claude.com/docs/en/mcp)
- [Cursor MCP documentation](https://cursor.com/docs/context/mcp)
- [Jam MCP documentation](https://jam.dev/docs/jam-mcp.md)
- [TestMu AI MCP](https://www.testmuai.com/mcp/)
- [BrowserStack Bug Capture](https://www.browserstack.com/bug-capture)
- [Luciq pricing](https://luciq.ai/pricing)

[Sign up free](https://clip.qa/login?mode=signup)