Bug bounty · public · paid

We want you to try to break us.

AWAIS defends 66 rules across 261 attack patterns. If you can break it, we want to know.

Submit a finding →Bounty matrix ↓Red-team challenge →

$250 · $1,000 · $5,000 · $10,000 · 72h triage

Bounty matrix

Severity in. Dollars out.

Same matrix published on /security and written into our MSA. Bounty is pre-approved at triage, paid on fix verification.

Severity

Amount

Examples

Low

$250

Info disclosure with no PII. Missing CSRF on a low-risk action. Non-sensitive misconfiguration.

Medium

$1,000

Stored XSS. IDOR with bounded scope. Auth bypass to a non-admin account. Rate-limit bypass on a sensitive endpoint.

High

$5,000

Cross-tenant data access. AWAIS bypass leading to PII exposure. MFA bypass. Pre-auth SSRF reaching internal services.

Critical

$10,000

Full account takeover with no user interaction. Mesh-federation compromise. Multi-tenant privilege escalation. Pre-auth RCE.

Severity assigned by Gladius security against the matrix above. Disputes go to a written second-look — we will explain our reasoning and we will listen to yours.

Service-level agreement

What we owe you, and when.

Initial response

72 hours

A human reads your report and confirms receipt. Not an autoresponder — an engineer with context on the affected surface.

Severity triage

7 days

Reproduced or rejected, with severity assigned against the matrix above and a written rationale. Bounty pre-approved at triage if accepted.

Fix or status update

30 days

Either shipped + verified, or a written status update with the next milestone date. Critical findings get a public incident card on /awais/incidents after fix.

Scope

The fence, drawn explicitly.

Anything below the in-scope list is fair game. Anything in the out-of-scope list will not be paid and may not be indemnified. When in doubt, email first.

In scope

Test these surfaces.

*.gladiuscrm.com — marketing, app, founders portal, all subdomains
Sister verticals — gladiusbdc.com, gladiusturf.com, gladiusstone.com, and their auth subdomains
Public API endpoints — /api/webhooks/adf, /api/health, /awais/api/status, /awais/api/incidents, /awais/api/badge
AWAIS itself — rule evasion, signature bypass, mesh-federation forgery, deception detection
Auth surfaces — homegrown TOTP MFA, session cookie, magic links, the accept-invite flow

Out of scope

Don’t test these.

Social engineering of Gladius staff, dealers, or partners — phishing, vishing, smishing
Physical attacks against Gladius offices, data centers, or staff devices
Volumetric DoS / DDoS — that is CDN territory, not AWAIS
Attacks against named pilot dealer accounts in production — we will provision a test tenant on request
Findings against third-party vendors (Stripe, Twilio, Resend, Clerk-legacy) without a Gladius-specific impact path
Self-XSS, missing security headers with no exploit, theoretical issues without a working PoC

Email security@gladiuscrm.com for a provisioned test tenant.

The red-team challenge

30 days. One sandbox. $15,000 best-in-show.

Formal program for qualified firms and accomplished solo researchers. Tighter scope, dedicated test tenant on the live federation, and a guaranteed top prize on top of every accepted bounty.

Format

Invited engagement — apply via security@

Sandbox

Dedicated test tenant on the live federation

Window

30 days of monitored access

Best in show

$15,000 + hall-of-fame top slot

Stacking

All findings also paid against the bounty matrix

Disclosure

Joint exec summary published if you let us — your call

Apply for the challenge →See the hall of fame ↓

Rules of engagement

Five lines. Don’t cross them.

Follow these and you are protected under our safe-harbor clause. Step over and we cannot defend you legally.

01Test only against your own tenant. Cross-tenant testing requires written approval.
02No real PII exfiltration. Demonstrate access with synthetic data; do not extract production PII.
03Keep request rates reasonable. Pause and email us if you trip a rate-limit you didn't intend.
04Written disclosure within 30 days. Coordinated public disclosure after the fix ships.
05Good-faith research only. Findings used to extort, leak, or trade get prosecuted, not paid.

How to submit

One email. Five fields.

security@gladiuscrm.com

Subject: Short title

Vector: Attack class

Impact: What you can reach

Repro: Step-by-step

Proof: PoC code, screenshot, or HAR

Open my mail client →

Hall of fame

The wall is empty. Be #1.

No fakes, no seeded names, no “internal team” placeholders. Every slot here will be a real researcher who broke something real.

Reporter #001

First reporter goes here.

The first researcher to land an accepted finding gets the top slot and a permanent mention on /awais/researchers when that page goes live.

What AWAIS does · and what it doesn’t

The defender, honestly described.

Know what AWAIS is before you start. Knowing what it isn’t is half the bounty.

AWAIS defends

Defends SQL injection, XSS, SSRF, RCE, and prototype pollution at the app edge.
Defends auth bypass, JWT tampering, session fixation, and credential stuffing.
Defends rate abuse, scraping, account enumeration, and inventory probing.
Defends tenant-isolation boundaries on every tRPC handler.
Defends accidental PII egress with response-body scanners and encrypted-column read-path enforcement.

AWAIS does not

Does not replace your CDN WAF — Cloudflare / AWS WAF still absorb volumetric layer-3/4.
Does not act as a SIEM — Splunk / Sumo / Elastic still own the log warehouse.
Does not pen-test itself — AWAIS scores via a deterministic synthetic adversary; humans (you) close the loop.

Full technical reference at /awais/playbook.