Every bug arrives in Jira with video, console logs, AI repro steps, root cause, a suggested code fix, and a regression test. MCP Server feeds the full context to Cursor, Claude Code, and Windsurf.
Works inside Jira and GitHub. MCP connects to Cursor, Claude Code, Windsurf. 14-day free trial.
AI Solution Finder
✦ Root Cause
TypeError at checkout.js:142
user object is null because payment
was initiated before login completed
✦ Suggested Fix
if (!user) {
await waitForAuth();
}
// Add null check before payment init
✦ Confidence: High (92%)
✦ Regression test: Generated ✓
Bug report says “it doesn’t work.” No steps, no logs, no context. You spend 30 minutes trying to reproduce.
Every bug arrives with video, console logs, network errors, numbered repro steps, and the root cause identified by AI.
No regression test. Same bug resurfaces 3 weeks later. Another fire drill.
AI auto-generates a Playwright/Cypress regression test from the recording. Push to CI. That bug never comes back.
You manually copy logs into AI tools, losing context. The fix suggestion is generic and often wrong.
MCP Server delivers full bug context to Cursor/Claude Code. Say “fix this bug” and AI has everything it needs.
Go beyond “here’s what broke” to “here’s how to fix it.” The AI Solution Finder analyzes console errors, network failures, and stack traces to generate root causes and code fixes with confidence levels.
solution-finder.output
✦ Root Cause
File: src/checkout.js
Line: 142
user.paymentMethod is undefined
Race condition: payment init before auth
✦ Suggested Fix
const user = await waitForAuth();
if (user?.paymentMethod) {
initPayment(user.paymentMethod);
}
✦ Confidence: High (92%)
✦ Similar fixes: 3 resolved issues
After a bug is captured, AI generates a regression test from the recording data: the exact navigation steps, inputs, and expected outcomes. Copy, download, or push directly to your repo.
checkout-payment.spec.ts
describe('Checkout Payment Flow', () => {
it('handles unauthenticated user', async () => {
// Navigate to checkout
await page.goto('/checkout');
// Attempt payment without login
await page.goto('#pay-btn');
// Verify redirect to login
await expect(page.url())
.toContain('/login');
// Verify redirect to login
await expect(page).not.toThrow();
});
});
The MCP Server exposes Sniffer’s full bug data to AI coding tools using the open Model Context Protocol (MCP) standard. Cursor, Claude Code, and Windsurf connect natively. Engineers say “fix this bug” and the AI pulls video context, logs, repro steps, and the suggested fix automatically.
npx @sniffer/mcp-server init
# Add to .cursor/mcp.json or claude_desktop_config.json:
{
"mcpServers": {
"sniffer": {
"command": "npx",
"args": ["@sniffer/mcp-server"],
"env": { "SNIFFER_API_KEY": "your-key" }
}
}
}
cursor / claude code
> fix the checkout bug from Sniffer
✦ Fetching from Sniffer MCP...
get_bug_report(id: "BUG-4821")
video, console, network, repro steps
AI solution + regression test
✦ Analysis complete.
Applying fix to checkout.js:142...
Running generated regression test...
✓ Fix applied. All tests passing.
✓ Regression test added to suite.
Smart Error Grouping detects when multiple customers hit the same bug across all three capture channels. Trending Issues dashboard ranks bugs by customer impact so you fix what matters first.
Affecting 23 customers across 4 browsers. First seen 3 days ago. Trending ↑
12 customers. 502 Bad Gateway. Spike correlates with deploy #4821.
5 customers on Safari. Appears after selecting 10+ filter options.
2 customers on Firefox. Intermittent. Low priority.
No need for a separate project management tool for bug tracking. Sniffer’s built-in Kanban board tracks bugs from capture to resolution with severity, assignees, tags, and estimation. One-click escalate to Jira or Asana when needed.
To Do 3
In Progress 2
Done 4
Install the Sniffer MCP server (one command). Cursor, Claude Code, and Windsurf connect natively. Say “fix the checkout bug from Sniffer” and AI receives the full context: video, logs, repro steps, root cause, and suggested fix.
Playwright and Cypress. Tests are generated from actual recording data (real navigation, real inputs, real assertions). You can edit before exporting. Push directly to GitHub or GitLab.
Every suggestion comes with a confidence score. High-confidence fixes (80%+) are generated from real stack traces and network errors, not generic patterns. Engineers always review before applying.
Jira, Asana, Azure DevOps, GitHub Issues, and Linear. Bugs arrive with the full AI report attached. One-click create from the Sniffer dashboard or auto-create from the support workflow.
Sentry and LogRocket monitor production errors and sessions. Sniffer captures bugs with full context from three channels (extension, recording links, JS snippet), then AI generates repro steps, root cause, fix, and regression test. They show you the problem. Sniffer helps you solve it.
14-day free trial. No credit card. Full AI pipeline from day one.