AXguard
AXguard is a pre-ship security gate — a CLI and AI-agent plugin — that scans vibe-coded and AI-built applications for secrets, broken auth, injection, SSRF, XSS, cloud misconfiguration and AI-agent/LLM risks before you ship.
Who it is for
Builders shipping with Claude Code, Cursor and similar agents; AppSec and platform teams who need a fast gate on AI-generated code; anyone who wants an open scanner rather than another SaaS trial.
Problems it addresses
- AI coding agents produce working apps faster than review can keep up
- Secrets and auth mistakes slip into vibe-coded projects before a first security pass
- LLM and agent integrations introduce prompt-injection and tool-abuse paths most SAST never saw
- Teams need a local gate they can put in CI without inventing a full pentest programme
How it helps
- Ships as CLI axguard plus agent plugins for Claude Code, Cursor and similar tools
- Includes MCP support for agent workflows
- Uses deterministic rule packs covering secrets, broken auth, injection, SSRF, XSS, cloud misconfiguration and AI-agent/LLM checks
- Supports CI gating with --fail-on so merges can stop on matched findings
Limitations
- It is a scanner and gate, not a substitute for a scoped penetration test
- Findings still need human triage; false positives and false negatives both exist
- Coverage grows with the project — check the repository README for the current rule set
Project facts
| Repository | github.com/Awarexone/AXguard |
|---|---|
| Language | Python |
| Status | Live |
| Licence posture | Permissive open source (see repository) |
| Telemetry | None claimed in project positioning; confirm in the repository |
Related services and research
- Cybersecurity services
- Open source program
- AI voice cloning attacks: how they work and how to stop them
Frequently asked questions
What is AXguard for?
Pre-ship review of applications built quickly with AI coding agents: catch secrets, auth and injection issues, plus AI-agent/LLM risks, before the code goes live.
Is AXguard a full penetration-testing platform?
No. The project is explicit that it is a pre-ship security gate, not a full pentest platform.
Is AXguard a replacement for penetration testing?
No. Use it as an early gate. Authorised penetration testing and adversarial review still matter for high-risk systems.
Get the code
Installation, current capabilities and licence terms live in the repository. The tool is free; scoped human-led testing is a separate engagement.