Type to search across all documentation

Documentation

Savants has two products. Pick your path.

AI Agent Guardrails

Prevent your AI agent from deleting databases, overspending, or doing anything you didn't authorize. Rules defined in plain English, enforced in code. Works with any LLM.

npm install @savants.dev/guard
Quick Start Install, write a guard rule, see it block an action. 2 minutes. Recipes Copy-paste patterns: Vercel AI SDK, spend limits, deploy safety. Guard DSL Reference All 16 operators: eq, gt, contains, matches, and more. Playground Write and run guard rules live in your browser.

Code Intelligence CLI

Give your LLM full codebase context. Semantic search, error diagnosis, blast radius analysis. Installs as an MCP server for Claude Code, Cursor, and Windsurf.

curl -fsSL savants.sh | sh
Install Guide One command. Auto-detects repos, clusters, services. CLI Reference All commands: diagnose, search, callers, blast-radius.

Connect to Cloud (optional)

When your team needs central rule management, live updates without redeploying, and a dashboard showing what was blocked:

1 Sign up Google or GitHub OAuth
2 Get API key From your dashboard
3 Add managed: true To your createGuard call

Advanced

FSM Schema Format Define state machines for workflow management and monitoring. Cloud API Reference REST endpoints for managed rules, schemas, and the registry.

Resources