# Capsule Security

> Capsule Security is an AI agent runtime security platform that continuously monitors AI agent behavior and intervenes during runtime at the first sign of anomalous or unsafe activity. It detects and blocks risky commands, unsafe tool usage, sensitive data exposure, and unintended action chains before they execute — without modifying agent code, frameworks, or workflows.

Capsule operates as an independent, agentless security layer for any AI agent that takes actions, including coding agents (Cursor, Claude Code, GitHub Copilot), enterprise SaaS agents (Microsoft Copilot Studio, Salesforce Agentforce, ChatGPT Enterprise, ServiceNow, Atlassian), and home-grown agents built on AWS Bedrock, Azure AI Foundry, and GCP Vertex. The platform was founded by security veterans Naor Paz (CEO) and Lidan Hazout (CTO), is backed by Forgepoint Capital and Lama Partners, raised a $7M round, and is SOC 2 certified.

## Core platform

- [Homepage](https://www.capsulesecurity.io/): Platform overview covering runtime monitoring, runtime intervention, and frictionless agentless deployment for AI agents.
- [About](https://www.capsulesecurity.io/about-us): Company philosophy, founders (Naor Paz, Lidan Hazout), advisors (Chris Krebs, Jim Routh, Mandy Andress, Omer Grossman, and others), and investors.
- [Blog](https://www.capsulesecurity.io/blog): Security research, vulnerability disclosures, and product announcements from the Capsule team.
- [Get a demo](https://calendly.com/naor-capsule/30min): Schedule a 30-minute platform demo with the founding team.

## Product capabilities

- **Guardian Agent**: Always-on enterprise guardian that continuously discovers, observes, and secures AI agents while detecting threats and suspicious behavior in real time.
- **Frictionless Discovery**: Agentless integration that automatically discovers AI agents across home-grown systems, SaaS agent platforms, and endpoint environments.
- **Agent Security Graph**: Maps how agents think, act, and interact at runtime by analyzing relationships between agents, tools, data, and actions to reveal risky paths and control gaps.
- **Deep Observability**: Real-time visibility into agent actions, decisions, and execution paths for investigation, governance, and safe scaling.
- **Runtime Protection**: Enforces security and governance policies in real time, blocking unsafe or risky agent behavior before actions execute.
- **Agent Identity Control / Agentic Policy Control**: Maintains clear ownership, least privilege, and accountability across agent environments.
- **Whitebox Red Teaming**: Proactively uncovers weaknesses in agent logic, prompts, and behaviors, feeding insights directly into runtime protection.

## Supported integrations

- **AI agent builder platforms**: AWS Bedrock, Azure AI Foundry, GCP Vertex AI.
- **AI coding agents**: Claude Code, Cursor, GitHub Copilot, and other MCP-connected developer tools.
- **Enterprise AI agents**: ChatGPT Enterprise, Microsoft Copilot Studio, Salesforce Agentforce, ServiceNow, Atlassian.

## Security research and blog posts

- [The Rise of Guardian Agents: Securing the Agentic AI Ecosystem](https://www.capsulesecurity.io/blog-post/the-rise-of-guardian-agents-securing-the-agentic-ai-ecosystem): The case for a guardian-agent approach to securing agentic AI.
- [CurseChain: How Hidden README Comments Trick Cursor Into Stealing — and Spreading — Your SSH Keys](https://www.capsulesecurity.io/blog-post/cursechain): Disclosed attack chain abusing hidden README content in Cursor IDE.
- [PipeLeak: The Lead That Stole Your Database — Exploiting Salesforce Agentforce With Indirect Prompt Injection](https://www.capsulesecurity.io/blog-post/pipeleak-the-lead-that-stole-your-database-exploiting-salesforce-agentforce-with-indirect-prompt-injection): Indirect prompt injection vulnerability in Salesforce Agentforce.
- [ShareLeak: Taking the Wheel of Microsoft's Copilot Studio (CVE-2026-21520)](https://www.capsulesecurity.io/blog-post/shareleak-taking-the-wheel-of-microsofts-copilot-studio-cve-2026-21520): CVE disclosure for Microsoft Copilot Studio.
- [Why MCP Gateways are a Bad Idea (and What to Do Instead)](https://www.capsulesecurity.io/blog-post/why-mcp-gateways-are-a-bad-idea-and-what-to-do-instead): Architectural critique of MCP gateway patterns and recommended alternatives.
- [ClawGuard: Open Source Security for the Agentic Era](https://www.capsulesecurity.io/blog-post/clawguard-open-source-security-for-the-agentic-era): Open-source agent security plugin from Capsule.
- [Capsule Security Raises $7M to Prevent AI Agents from Going Rogue in Runtime: Intent is the New Perimeter](https://www.capsulesecurity.io/blog-post/intent-is-the-new-perimeter): Funding announcement and the "intent is the new perimeter" thesis.

## Optional

- [Privacy Policy](https://www.capsulesecurity.io/policy-privacy)
- [Terms & Conditions](https://www.capsulesecurity.io/terms-of-use)
- [LinkedIn](https://www.linkedin.com/company/capsule-security/)