Real user sessions.AI-written bug reports.GitHub issues in one click.
Vigil watches every user session, clusters repeated failures, and hands you a prioritized queue of developer-ready bug reports — before your users open a support ticket.
VIGIL
Where every broken session becomes a report,and every report becomes a fix.
SIGNAL
Through the noise of real user behavior,the failures that matter rise first.
TRIAGE
Before your users open a ticket,the bug is already in your queue.
Four steps from broken session to closed issue.
Record
Install @vigil/sdk with a single script tag or npm package. It captures DOM mutations, clicks, errors, network failures, and navigations using rrweb — with all inputs masked by default.
Ingest
Sessions are batched and flushed to Vigil's ingest API. Storage is asynchronous and non-blocking. Your app never feels the overhead.
Triage
Every completed session enters the AI triage pipeline. Claude analyzes the full event timeline, detects what went wrong, and decides whether it is a new issue, a duplicate of a known failure, or normal behavior.
Act
Work from a prioritized issue queue. Each issue has a root cause, a suggested fix, reproduction steps, and evidence. Raise a pre-filled GitHub issue with one click.
AI that reasons, not just flags.
Most tools detect rage clicks and JS errors. Vigil understands what the user was trying to do, whether they succeeded, and exactly why they failed.
Every session gets a friction score, a goal completion verdict, and a list of issues with root causes and suggested fixes — written by Claude, ready for your team.
{
"session_summary": "Payment failed on POST /api/payment 500",
"goal_completed": false,
"friction_score": 92,
"issues": [
{
"title": "Payment fails silently on 500 error",
"severity": "P0",
"confidence": 0.91,
"root_cause": "POST /api/payment 500; UI lacks error state",
"suggested_fix": "Handle !res.ok and surface toast notification",
"reproduction_steps": [
"Open /checkout",
"Click pay button",
"Observe POST /api/payment -> 500",
"Observe no user-facing error shown"
]
}
]
}Everything your team needs to ship without breaking things.
Issue Grouping
Vigil clusters repeated failures automatically. Instead of 500 separate reports, you get one issue affecting 500 sessions — with the first seen, last seen, and all the evidence.
Friction Score
Every session gets a 0–100 friction score. Sort your queue by who struggled most, not just who errored out.
Reproduction Steps
AI generates step-by-step reproduction paths from the session timeline. Paste them straight into your GitHub issue.
GitHub Integration
One-click issue creation with AI-written body, severity labels, session replay links, and stack traces pre-attached. One issue per bug group — not one per affected user.
Auto-Raise Mode
Configure Vigil to automatically raise GitHub issues for P0/P1 bugs above your confidence threshold. Your repo gets the issue before anyone on your team notices.
Privacy First
All input values are masked by default. Passwords are never captured. The AI only sees structured event summaries — never raw replay data or unmasked text.
The issue queue developers actually want.
| Severity | Issue Summary | Affected Users | Confidence | Last Seen | Status |
|---|---|---|---|---|---|
| P0 | Payment submission fails with 500 error | 500 sessions | 91% confidence | 2h ago | GitHub #419 |
| P1 | Auth token expires mid-session without refresh | 134 sessions | 87% confidence | 5h ago | |
| P1 | Checkout form submit button unresponsive on mobile | 89 sessions | 83% confidence | 1d ago | vigil-auto-raised |
| P2 | Confusing error message on invalid promo code | 41 sessions | 74% confidence | 2d ago | |
| P3 | Slow navigation from /dashboard to /settings | 12 sessions | 61% confidence | 3d ago |
Sort by severity. Filter by environment. Raise a GitHub issue without leaving the page.
Vigil vs. watching recordings yourself.
Without Vigil
- Watch recordings manually for hours
- Guess which sessions are worth your time
- Write bug reports from scratch
- Miss duplicate bugs filed separately
- Find out when users complain on Twitter
With Vigil
- AI triages every session automatically
- Issues ranked by severity and affected sessions
- Root cause, fix, and reproduction steps written for you
- Repeated failures clustered into one issue group
- Bug is in your queue before users notice
What developers say.
“We had a payment failure affecting 12% of checkout attempts for three days. Nobody noticed because there was no JS error in Sentry — just a silent 500. Vigil caught it on day one.”
Pushpendra M.
Product Design Intern, Someone
“The GitHub integration is the thing. One click and there's a proper issue in the repo with the stack trace, the replay link, and steps to reproduce. Our PMs can even understand them.”
Sachin V.
Founder, Acadable Labs
Drop in. Done in two minutes.
<script src="https://cdn.vigil.dev/sdk.js" defer></script>
<script>
Vigil.init({
projectKey: "pk_playground",
environment: "production",
release: "1.0.0", // Replace with your release/version
});
</script>Stop watching replays. Start fixing bugs.
Vigil triages every session automatically. Your first 1,000 sessions are free.
No credit card required. Installs in under 2 minutes.