AgentTrust

AgentTrust: Security gateway and policy enforcement system for autonomous AI agent runtimes.

AgentTrust is a security gateway and policy enforcement system designed for autonomous AI agent runtimes like LangChain, CrewAI, AutoGen, and OpenAI Assistants. It establishes zero-trust capability boundaries, intercepts tool invocations before execution, mandates human approvals on high-blast actions, and maintains an immutable audit ledger. The system evaluates the real-time reversibility and scope of actions, detects goal drift and semantic hijack attempts, and enforces strict downstream capability downscoping. It provides transparent egress proxy filtering to block server-side request forgery alongside post-quantum cryptographic attestation. Commercial licensing and enterprise support are available, with evaluation under the MIT License.

AgentTrust at a glance

Best for: Software engineer
Runs on: API, self-hosted
Pricing: Evaluation and Research under MIT License; Commercial deployment via inquiry.
Released: October 2026
Category: AI agent tool
Website: github.com/srn0129/agenttrust

Key features

Deny-by-default policy engine
Cascading delegation and downscoping
Semantic blast radius analysis
Human-in-the-loop approvals
Verifiable audit ledger
Transparent egress proxy
Goal drift detection
Emergency air-gap killswitch

Use it for

Enforcing zero-trust security policies for autonomous agent runtimes
Requiring human approvals for high-impact tool executions
Maintaining verifiable, hash-linked cryptographic audit ledgers
Blocking SSRF and unauthorized outbound network calls from agents

Alternatives to AgentTrust

Agent Delegated Authority — It defines cryptographically verifiable authorization receipts and deterministic boundaries.
Password Agent Sys — Enforces strict permission control and cryptographic task delegation for multi-agent systems.
iFixAi — Audits agent behavior independently without building manual eval harnesses.
AgentDock — It coordinates multi-device workflows and provides secure environment access.
Aegis Kernel — Stops prompt injection and tool-hijacking at the OS syscall boundary.
AgentGate — Blocks attacks in under 35ms with a local neural classification head.
Agent Governance Toolkit — Intercepts tool calls and policy enforcement in deterministic application code.
Agent Vault — Protects real cards and passwords from prompt injections by using mock tokens.
TaintWall — Enforces deterministic security policies at the tool boundary for AI agents.
Exo Harness — Separates trusted state and security from agent logic for safe execution.
AgentGuard — Get a security and quality assessment of AI skills in seconds without an account.
Composio — Equips AI agents with tools, managed user auth, and sandboxes across 1,500+ apps

All AgentTrust alternatives