Skip to main content

​
ABS Core

ABS Core is a real-time governance engine for AI agents. It monitors every action an AI agent takes, enforces security policies, and produces an immutable audit trail.

Quick Start

Get protected in 5 minutes

API Reference

Full API documentation

Policy Engine

Write custom security policies

SDK

Integrate with any agent framework

​
Why ABS Core?

AI agents are increasingly autonomous — writing code, making API calls, accessing databases. Without governance, every agent is a liability.
ProblemABS Solution
Agent sends PII to external APIPolicy Engine blocks in <5ms
Agent credentials leaked in logsSecret Vault injects at runtime (JIT)
Agent stops respondingHeartbeat Sentinel detects in 5 minutes
Compliance audit requiredImmutable Ledger with hash chain
New attack vector discoveredCHI classifies intent semantically

​
Architecture

Agent (Cursor, Copilot, etc.)
  │
  ▼
┌─────────────────────────────────┐
│         ABS Shield (Proxy)      │
│  ┌──────────┐  ┌─────────────┐  │
│  │ Secret   │  │ Rate        │  │
│  │ Vault    │  │ Limiter     │  │
│  └──────────┘  └─────────────┘  │
│  ┌──────────┐  ┌─────────────┐  │
│  │ Policy   │  │ CHI         │  │
│  │ Engine   │  │ (Semantic)  │  │
│  └──────────┘  └─────────────┘  │
│  ┌──────────┐  ┌─────────────┐  │
│  │ Heartbeat│  │ Audit       │  │
│  │ Sentinel │  │ Ledger      │  │
│  └──────────┘  └─────────────┘  │
└─────────────────────────────────┘
  │
  ▼
LLM Provider (OpenAI, Gemini, Anthropic)