Elemm Guardian
The zero-trust security layer built natively into the Elemm Gateway. It acts as an intelligent API gatekeeper, continuously inspecting every tool call, validating nested arguments, and enforcing strict access policies before requests ever reach your backend infrastructure.
Zero-Trust Whitelisting
By default, the Elemm Gateway operates in a Zero-Trust mode. Unless explicitly whitelisted, all AI agent requests to specific landmarks or API actions are instantly denied. This strict boundary control ensures that only pre-approved routes are accessible.
Deep Argument Inspection
The Guardian doesn't just look at the requested endpoint. It performs a recursive scan on nested JSON payloads, lists, and parameter maps. If any deeply nested argument contains a forbidden pattern, the entire request is blocked before hitting the API.
Pattern & Exploit Blocking
Protect your infrastructure against prompt injections and malicious inputs using simple keywords or compiled Regular Expressions. Block dangerous commands like 'drop table', 'delete', or 'purge' proactively.
HTTP Method Enforcement
You can restrict which HTTP methods an AI agent is permitted to use globally. For instance, you can easily configure the policy to only allow safe GET requests, blocking any state-mutating POST, PUT, or DELETE operations.
Anti-Injection Query Protection
When agents search for new landmarks dynamically, their search queries are parsed, cleaned, and validated. This mitigates regex-injection attempts and prevents agents from obfuscating searches to discover restricted API areas.
Contextual Remedies
When a request is blocked, the Guardian doesn't just fail silently. It returns custom-configured 'remedies'—contextual hints that guide the autonomous agent back to the correct path, reducing hallucination loops and errors.
Ready to Secure Your Agents?
Dive into the technical documentation to learn how to configure the Guardian layer, define whitelists, and protect your endpoints.
Read Security Docs