Agent Security Indexbeta
Home / Agent runtime security & posture / Obsidian AI Agent Security

Obsidian AI Agent Security

Quick answer: Obsidian AI Agent Security is a commercial agent runtime security and posture platform from Obsidian Security. Obsidian Security's AI Agent Security platform discovers AI agents across the enterprise, maps each one to its owner, tools, MCP servers and models, right-sizes permissions, and enforces policy at runtime. A Runtime Policy Decision Engine evaluates each agent action with identity, application and customer-provided context and returns allow, block, or require approval where supported. Enforcement points include a sidecar for coding agents, an MCP Security Proxy that checks tool calls before they reach supported apps, and a Policy Decision API for customer-operated gateways. The MCP Security Proxy launched Nov 10, 2026 as a governed checkpoint between AI clients and supported MCP apps: it identifies the user, routes to the correct tenant, keeps the app token out of the AI client, and checks each tool call against the same policies. The product is sold as commercial SaaS with regional hosting. A Free plan is listed at $0 per month for up to 1,000 users for discovery of unsanctioned AI and shadow SaaS; Foundations and Advanced modules, which include agent governance and runtime guardrails, have no public list price.
Commercial
Visit site ↗
Obsidian AI Agent Security at a glance
CategoryAgent runtime security & posture
Official sitewww.obsidiansecurity.com/ai-agent-security
MakerObsidian Security
Ownership / acquisitionNot published / not yet verified
License modelCommercial
LicenseNot published / not yet verified
Public pricingFree $0/month (up to 1,000 users); Foundations and Advanced by quote
Source codeNot published / not yet verified
Last verified

Facts are taken from the vendor's own site, GitHub, or a linked public source. Blank means we could not confirm it, not that it does not exist. Vendors can request a correction for free.

Lab result

Not yet tested. Lab testing is in preparation. No tool on the site has a lab result yet.

About planned lab testing

Similar tools

Pillar Security

Prisma AIRS

Cisco AI Defense

Frequently asked questions

What is Obsidian AI Agent Security?

Obsidian AI Agent Security is a commercial agent runtime security and posture platform from Obsidian Security. Obsidian Security's AI Agent Security platform discovers AI agents across the enterprise, maps each one to its owner, tools, MCP servers and models, right-sizes permissions, and enforces policy at runtime. A Runtime Policy Decision Engine evaluates each agent action with identity, application and customer-provided context and returns allow, block, or require approval where supported. Enforcement points include a sidecar for coding agents, an MCP Security Proxy that checks tool calls before they reach supported apps, and a Policy Decision API for customer-operated gateways. The MCP Security Proxy launched Nov 10, 2026 as a governed checkpoint between AI clients and supported MCP apps: it identifies the user, routes to the correct tenant, keeps the app token out of the AI client, and checks each tool call against the same policies. The product is sold as commercial SaaS with regional hosting. A Free plan is listed at $0 per month for up to 1,000 users for discovery of unsanctioned AI and shadow SaaS; Foundations and Advanced modules, which include agent governance and runtime guardrails, have no public list price.

Is Obsidian AI Agent Security open source?

No. We list Obsidian AI Agent Security as a commercial product, and no public source repository is recorded for it.

How much does Obsidian AI Agent Security cost?

Published pricing: Free $0/month (up to 1,000 users); Foundations and Advanced by quote. Checked on November 10, 2026; confirm current prices on the official site.

Who makes Obsidian AI Agent Security?

Obsidian Security.

What are alternatives to Obsidian AI Agent Security?

Other agent runtime security and posture platforms listed here include Pillar Security, Prisma AIRS, Cisco AI Defense, HiddenLayer and Zenity. See all 23 in Agent runtime security & posture.