Salus

Developer tool

Runtime authorization layer that wraps the tool an agent uses to write to a system, holds an action that lacks evidence before the provider is called, returns the exact missing step to the agent and records a receipt for every decision; works with OpenAI, Anthropic, LangChain, LangGraph, CrewAI, AutoGen, Vapi, Retell and MCP; Pilot plan $500 a month and Partner plan $2,500 a month, with a custom Enterprise plan, priced by agents and action checks and not by seat.

Find alternatives to Salus

TypeDeveloper tool

RoleStandalone tool

Bought forWatch an AI agent for suspicious or runaway behaviour, Build and run AI agents

AvailabilitySold

DeploymentSaaS

Intended forDeveloper, Enterprise

SecurityNot recorded

StatusActive

Sold withinSold on its own

PricingSubscription, Enterprise quote — official pricing ↗

Pricing, Deployment, Sold within — usesalus.ai · 20 Oct 2026

What it does

Guardrails and safety
Agent orchestration

Features 4

PreventHolds an unsupported write before the provider is called.
RepairReturns fixable actions to the agent with the missing evidence and the next step.
ProveShows the policy, evidence, timing, verdict and execution result for every decision.
Shadow and enforcement modesEach protected route can run in shadow or enforcement mode, chosen per route.

usesalus.ai

Sources

Pricing

Pricing page · 20 Oct 2026

Pricing page: Pilot $500 / month (10 agents, 200K action checks a month), Partner $2,500 / month (50 agents, 1.5M action checks a month), Enterprise custom; unlimited seats on every plan.

Description

Official documentation · 20 Oct 2026

Home page: "Salus holds unsupported writes before your backend performs them."

What it does

Official documentation · 20 Oct 2026

Home page: "Salus holds unsupported writes before your backend performs them."

Sold within