Documentation
Learn how to set up, configure, and integrate Aguardic into your workflows.
Welcome to Aguardic
Aguardic is an AI governance platform that evaluates every AI agent action, pull request, LLM response, and document against your policies — blocking violations, enforcing rules, and generating audit-ready evidence automatically.
Three-Layer Governance Engine
Aguardic uses a three-layer evaluation engine:
Deterministic Rules
Fast, field-level conditions with operators (contains, equals, greater than, regex match). Evaluated in milliseconds.
Semantic AI
LLM-powered evaluation for nuanced checks that rules alone can't handle (tone, intent, context).
Custom Knowledge (RAG)
Upload your organization's documents and evaluate against your specific context using vector search.
What you can govern
Code & Pull Requests
PR reviews, branch protection, code scanning
AI Outputs
LLM request/response evaluation, prompt injection detection
Documents & Storage
File content scanning, sharing policies, PII detection
Email & Messaging
Channel monitoring, email compliance, attachment policies
AI Agents
Agent action governance, tool call evaluation, session tracking
Getting Started
New to Aguardic? Start with the Quickstart guide to create your first policy and run your first evaluation in under 5 minutes.
Quickstart
Create your first policy and run an evaluation in under 5 minutes.
Core Concepts
Understand policies, rules, evaluations, and violations.
Integrate with Your Stack
JavaScript / TypeScript SDK
Install @aguardic/sdk and start evaluating in 3 lines of code. Handles auth, response unwrapping, and types automatically.
REST API
Use the evaluate endpoint directly with any language or HTTP client.
MCP Server
Connect MCP-compatible AI agents (Claude Desktop, Cursor) for automatic governance.
Integrations
Connect GitHub, Slack, OpenAI, Gmail, and 12+ more platforms.
Govern Your AI Systems
Register every AI application your organization operates, classify risk, and enforce policies across all connected integrations.
AI Systems
Register AI applications, auto-classify risk (EU AI Act aligned), and link integrations for unified governance.
Entity Governance
Govern AI decisions against individual context — patient medication lists, customer contracts, vendor SLAs.
Evaluation Sessions
Track multi-step agent workflows with context-aware policies that reason across an entire session.
Audit Trail
Full violation lifecycle — detection, investigation, resolution — with compliance-ready evidence.