MCP for AI Agents
Connect BugSnap to Claude Code, Claude Desktop, Cursor, or any MCP-compatible AI agent so it can read full bug context, including screenshots, with one tool call.
MCP for AI Agents
BugSnap captures already carry everything needed to fix a bug: the screenshot, console errors, failed network requests, environment, and the steps that led to it. The bugsnap-mcp server exposes that over the Model Context Protocol, so your AI agent can read a report instead of you pasting it in by hand.
1. Generate a BugSnap API key
Create a workspace API key in Settings: this is what authenticates the MCP server as you.
Open Settings → BugSnap API Keys2. Connect your AI tool
Pick your tool and paste the config. Replace bugsnap_xxx with the key you just generated.
claude mcp add --scope user -e BUGSNAP_API_KEY=bugsnap_xxx --transport stdio bugsnap -- npx -y bugsnap-mcp
Writes to your user-level config, available in every project. Use --scope project instead to share it with your team via .mcp.json.
{
"mcpServers": {
"bugsnap": {
"command": "npx",
"args": ["-y", "bugsnap-mcp"],
"env": {
"BUGSNAP_API_KEY": "bugsnap_xxx"
}
}
}
}Paste into claude_desktop_config.json, then fully restart Claude Desktop. macOS: ~/Library/Application Support/Claude/claude_desktop_config.json Windows: %APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"bugsnap": {
"command": "npx",
"args": ["-y", "bugsnap-mcp"],
"env": {
"BUGSNAP_API_KEY": "bugsnap_xxx"
}
}
}
}Paste into .cursor/mcp.json (project) or your global Cursor MCP settings.
{
"mcpServers": {
"bugsnap": {
"command": "npx",
"args": ["-y", "bugsnap-mcp"],
"env": {
"BUGSNAP_API_KEY": "bugsnap_xxx"
}
}
}
}Most MCP clients (Windsurf, Cline, VS Code, ...) read the same mcpServers shape from their own config file.
3. Ask your agent
Once connected, just point your agent at a bug.
Available tools
What the AI agent can call once connected.
bugsnap_list_capturesLists recent bug reports for your workspace (id, title, status, tag, created date).
bugsnap_get_captureFull context for one report: description, environment, console errors, network failures, user steps, and the screenshot itself. Accepts a capture id or a full share URL.
Test it without an AI client
Run the server directly with the official MCP Inspector to see both tools and try a real call before wiring it into your editor.
BUGSNAP_API_KEY=bugsnap_xxx npx @modelcontextprotocol/inspector npx -y bugsnap-mcp