Skip to main content

Gatekeeper

1. What is this?

An infrastructure security scanner for self-hosted AI APIs.

Point it at your project directory. It finds the deployment mistakes that your model scanner can't see: hardcoded API keys, missing rate limits, CORS misconfiguration, containers running as root, TLS certificates about to expire.

Who it's for: Developers running their own AI API gateway (LiteLLM, chinai-gateway, or any OpenAI-compatible endpoint). If you docker compose up an AI service, this is for you.

pip install gatekeeper-ai
gatekeeper audit -d /path/to/your-project

2. What problem does it solve?

You deployed an AI API. You tested the model — jailbreaks, prompt injection, the works. But did you check if your .env file is in .gitignore? Or if your API accepts requests without authentication? Or if your TLS certificate expires next week?

Model scanners test your model. Gatekeeper tests your deployment. They're different attack surfaces.

What you get:

  • A security score and a priority action plan — what to fix first, how long it takes
  • Plain-language risk explanations — not "CRITICAL severity", but "your API key is in git history and anyone who clones your repo can use it"
  • Copy-paste fixes — exact YAML snippets, exact commands
  • CI/CD integration — SARIF output, GitHub Actions, PR comments

3. Why not just use garak or promptfoo?

garak promptfoo Gatekeeper
What it tests Model responses Code data flow Deployment config
Finds hardcoded API keys? No No Yes
Finds missing rate limits? No No Yes
Finds expiring TLS certs? No No Yes
Finds CORS misconfiguration? No No Yes
Finds privileged containers? No No Yes
Backed by NVIDIA OpenAI (acquired) Independent (MIT)

garak and promptfoo are the right tools for model-level and code-level security. Gatekeeper is the missing layer: deployment-level. They don't compete — they stack.

4. Free or paid?

Free. MIT license. No API key required. No cloud service. No telemetry. Everything runs locally — your config files never leave your machine.

If you want deeper AI-powered analysis, bring your own DeepSeek/OpenAI key and we'll give you the prompt template. But the built-in knowledge base covers the common cases without any API calls.


Quick Start

pip install git+https://github.com/AAAjczz/gatekeeper.git

# Audit a project (file checks)
gatekeeper audit -d /path/to/chinai-gateway

# Full audit (file + network checks)
gatekeeper audit -d . -e https://api.deepseek.com/v1 -k YOUR_KEY

# Output formats
gatekeeper audit -f terminal   # default
gatekeeper audit -f sarif      # GitHub code scanning
gatekeeper audit -f html       # shareable report
gatekeeper audit -f all        # everything

Real Finding

Gatekeeper's multi-domain TLS probe scanned api.deepseek.com and found 3 different certificates across CDN nodes — earliest expiring in 8 days:

deepseek.io       → GoDaddy     → expires in 8 days
api.deepseek.com  → TrustAsia   → expires in 24 days
deepseek.com      → DigiCert    → expires in 177 days

No model scanner or code scanner would find this. Infrastructure auditing does.

License

MIT — github.com/AAAjczz/gatekeeper

Release files for gatekeeper-audit 1.3.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for gatekeeper-audit 1.3.0
File Size Uploaded
gatekeeper_audit-1.3.0.tar.gz 28.0 kB Details

Built distribution (wheel)

Table of built distributions (wheels) for gatekeeper-audit 1.3.0
File Interpreter ABI Platform
gatekeeper_audit-1.3.0-py3-none-any.whl Python 3 none any Details

Total release size: 55.8 kB

Release files / gatekeeper_audit-1.3.0.tar.gz

Download URL gatekeeper_audit-1.3.0.tar.gz
Size 28.0 kB
Tags Source
SHA-256 checksum
How to use checksums
27460972eed15333db9cfb751e03e3e43c3d15aacf84de23ea3fa6cc69a47019
BLAKE2b-256 checksum
How to use checksums
a6ea078db313f5a04284968c70fb9ab0361bc967f58205475913b48a3714f35f
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/6.2.0 CPython/3.14.5

Release files / gatekeeper_audit-1.3.0-py3-none-any.whl

Download URL gatekeeper_audit-1.3.0-py3-none-any.whl
Size 27.9 kB
Tags Python 3
SHA-256 checksum
How to use checksums
00761dacf44ee6799d805500582ad5967016fa1884131c120af6d80c4085aed6
BLAKE2b-256 checksum
How to use checksums
96b2cc2b941621ce52e51897b4b81be265e506e55cd69e8b4a78bb63e5a3884e
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/6.2.0 CPython/3.14.5

Release history Release notifications | RSS feed

This release

1.3.0 This release

2 release files

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page