OSS AI gateways ship with
known vulnerability patterns.

Open-source gateways move fast. Privilege isolation failures, key exposure paths, and multi-tenant boundary violations accumulate faster than security reviews can catch them. We've already found 600+ across the most-deployed gateways.

OSS velocity creates
an invisible attack surface

The same openness that makes OSS gateways attractive also means vulnerabilities accumulate in public, and every deployment configuration creates a new attack surface.

01

Releases outpace security reviews

Open-source AI gateways ship weekly. Each release brings new configuration surface, new API endpoints, and new trust boundaries — faster than any manual security review can track.

02

Custom configs create new attack paths

Every gateway deployment is configured differently — custom routing rules, provider fallbacks, multi-tenant key namespacing. The attack paths in your deployment are not the same as the ones in the upstream docs.

03

Findings surface post-breach, not pre-breach

Most teams discover gateway vulnerabilities when someone exploits them — a leaked API key, a prompt injection that reaches a privileged tool, or a multi-tenant data leak. At that point, the damage is done.

Continuous fuzz and probe,
before attackers do it for you.

Automated Vuln Discovery runs a corpus-driven audit agent against your gateway configuration — finding what standard CVE scanners miss, and generating regression fixtures so fixed vulnerabilities stay fixed.

2,900-Entry Corpus

Our Code-Audit LLM is fine-tuned on 2,900+ real-world LLM gateway vulnerabilities, NVD CVE data, and OWASP GenAI cases. It finds implementation flaws that static scanners and generic pentest tools miss because they don't understand gateway-specific attack patterns.

Runtime-Chain Audit

A second model analyzes live model requests, tool calls, and permission decisions to find policy bypass and privilege escalation in real traffic — the class of vulnerability that only appears under production load, not in static code review.

Adversarial Fixture CI

Every finding automatically generates a regression test fixture. Discovered vulnerabilities are added to a continuous CI suite — so when you upgrade your gateway version, you know immediately if a fixed vulnerability regressed.

Gateway-specific
vulnerability classes

We cover the vulnerability dimensions that matter for AI gateway deployments — not generic web security, but the patterns that show up specifically in LLM routing, multi-tenant key management, and tool-call authorization.

  • 1

    Privilege isolation failures — tenant A accessing tenant B's model keys or usage data.

  • 2

    Key exposure paths — provider API keys reachable via crafted requests or error responses.

  • 3

    Tool-call authorization bypass — skipping permission checks via prompt crafting or parameter manipulation.

  • 4

    Secrets in model context — credentials injected into retrieved context and returned in completions.

vuln scan · AI gateway deployment
# Scan report excerpt
target:   ai-gateway:4000
corpus:   v2.9 · 2,900 entries
scanned:  2026-06-04T09:12:00Z

FINDING [HIGH] CVE-class · Privilege Isolation
  Path: /key/generate · param: team_id
  Tenant A can generate keys scoped to Tenant B
  via unsanitized team_id in request body
  Fixture: fixtures/priv-isolation-001.json

FINDING [MED] Key Exposure · Error Response
  Path: /chat/completions · malformed model
  Stack trace leaks provider API key prefix
  Fixture: fixtures/key-exposure-007.json

Summary: 2 findings · 2 fixtures generated
proof_hash sha256:7c3d2a…  (Evidence-AIDR)

FAQ

Which gateways do you currently cover?

We cover the major open-source and cloud-native AI gateway deployments. We also validated corpus transferability against 300+ crypto wallet projects — the vulnerability patterns are consistent across trust-boundary enforcement in multi-tenant key management systems.

Does the scan require access to our production environment?

No. We can scan against a staging replica or a production-equivalent test environment. The audit agent needs network access to the gateway's API surface — not database access, not SSH access. Most teams point us at a dedicated staging deployment.

What do we get as deliverables?

A signed findings report with each vulnerability classified by severity, reproduction steps, and a generated adversarial test fixture. Findings are also logged as entries in the shared evidence chain — giving you a cryptographically verifiable record of your security posture at scan time.

Can we buy this standalone, without the Evidence-AIDR subscription?

Yes. Automated Vuln Discovery is purchasable independently. You get the private corpus, the audit agent, and adversarial fixtures — without requiring a SaaS subscription for the evidence aggregator.

What's in your
gateway deployment?

Point our audit agent at your gateway environment and we'll deliver a findings report with cryptographic evidence your team can verify offline.