Works with the AI Agents you already use
zeroarmor runtime defense
Protect AI agents with RASP-based defense against prompt injection, RCE, SSRF, and secret leaks.
Free
zeroarmor runtime defense
Example session with this skill installed
Run a security scan on my current directory to check for RCE and SSRF vulnerabilities.
- Read your context and instructions
- Compiled the zeroarmor runtime defense
I have performed a security diagnostic on the local directory using the ZeroArmor Titanium engine. The scan identified three high-risk vulnerabilities involving insecure system calls and unvalidated URL requests.
Security Audit Report
| Vector | Severity | Location | Description |
|---|---|---|---|
| Remote Code Execution | CRITICAL | app/executor.py:14 | Use of os.system() with unvalidated user input detected. |
| SSRF | HIGH | utils/fetcher.py:22 | Request to 169.254.169.254 detected without IP blacklisting. |
| Path Traversal | MEDIUM | api/files.py:45 | open() call uses raw string concatenation for file paths. |
| Secret Leak | HIGH | .env.example | Live Stripe Restricted Key found in example configuration file. |
Next steps
- Replace
os.system()with thesubprocessmodule usingshell=Falseand strict argument arrays. - Implement a URL validator that explicitly blocks the
169.254.0.0/16CIDR range. - Use
os.path.basename()or a secure path joining utility to sanitize file access.
Connects securely to your tools. The creator never sees your data.
About this skill
Vulnerable code and insecure LLM prompts expose AI agents to system-level compromise and data theft. This skill provides a real-time Runtime Application Self-Protection (RASP) layer to intercept exploits before they execute.
What it does
- Neutralize Jailbreaks identifies and blocks prompt injections, system prompt extraction, and persona tampering attempts.
- Intercept OWASP Top-10 traps SQL injection, Path Traversal, and Remote Code Execution attempts in Python codebases.
- Block SSRF prevents unauthorized probes to cloud metadata services like AWS, GCP, and Azure.
- Scrub Secrets monitors agent outputs to filter out leaked AWS keys, Stripe tokens, and JWTs in real-time.
- Automated Auditing scans local directories for existing security vulnerabilities and insecure coding patterns.
How it works
- Analyze Payloads using the
--checkcommand to evaluate specific code strings or prompt inputs for threats. - Scan Directories via the
--scancommand to identify vulnerabilities across an entire project. - Configure Enforcement by selecting
BLOCK,AUDIT, orDECEIVEmodes to control how the system responds to detected threats. - Monitor Outputs to ensure no sensitive credentials or system data are leaked to the end user.
Frameworks & tools
This skill is designed for Python codebases and AI agents. It integrates with developer tools like Cursor and Claude Code to provide DevSecOps coverage.
Why this beats prompting it yourself
Generic system prompts often fail against sophisticated prompt injections and DAN-style exploits. This skill uses a multi-vector defense engine that goes beyond text matching to identify execution-level risks and infrastructure probes.
Use cases
- Protect agents from prompt injection and system prompt extraction attacks.
- Secure production Python backends against RCE and SQL injection.
- Prevent cloud metadata theft in agents deployed on AWS, GCP, or Azure.
- Automatically scrub API keys from model responses before they reach users.
Known limitations
Requires a local Python environment to execute the ZeroArmor client. Focuses primarily on Python and LLM-specific vectors.
Requires local Python environment. Optimized for Python codebases and common cloud metadata endpoints.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 3 days ago
- Free with an Agensi account