# Prismor

> Prismor is a security control plane for AI agents. It intercepts every tool call an agent makes, enforces security policies at runtime, and logs a full audit trail in production.

Prismor governs what AI agents are allowed to do. It works as a gateway in front of Model Context Protocol (MCP) servers and agent frameworks, evaluating each tool call against policy as it happens, masking secrets before they leak, enforcing least-privilege and zero standing access, and discovering shadow AI usage. It integrates with MCP, the OpenAI Agents SDK, LangChain, and CrewAI, and with AI coding agents like Cursor, Windsurf, and Claude.

## Core concepts

- AI agent control plane: a single layer between agents and their tools where every action is governed.
- Runtime security for AI agents: policy is enforced at the moment an agent acts, not just when it is configured.
- Tool call interception: each call an agent makes is proxied, evaluated, allowed, blocked, or masked.
- Agent audit trail: a complete, queryable record of every action an agent took and why it was permitted.
- Least privilege and zero standing access: agents get only the access they need, only when they need it.
- Shadow AI discovery: surface unsanctioned AI agents and MCP servers operating in your environment.

## Docs

- [Documentation](https://www.prismor.dev/docs): Full product documentation
- [Claude Inference Hooks](https://www.prismor.dev/docs/claude-inference-hooks): Prismor as the AI security server for Claude Enterprise Inference hooks — allow/deny verdicts for claude.ai, Claude Code and Cowork prompts
- [MCP Integration](https://www.prismor.dev/docs/mcp): Model Context Protocol gateway for Cursor, Windsurf, Claude, and other AI assistants
- [Prismor](https://www.prismor.dev/docs/prismor): Open-source runtime security for AI coding agents
- [Sweep - Secret Scanner](https://www.prismor.dev/docs/sweep): Find and redact secrets leaked into AI tool configs
- [Cloak](https://www.prismor.dev/docs/cloak): Mask secret values from AI agents at runtime
- [AI Security](https://www.prismor.dev/docs/ai-security): Prompt injection detection and topic banning for agents
- [Policy](https://www.prismor.dev/docs/policy): Define agent policy as code, per user and per team
- [Quickstart](https://www.prismor.dev/docs/quickstart): Install, run the setup wizard, choose which rules block, and hook your agent
- [Runtime CLI Reference](https://www.prismor.dev/docs/cli-reference): Every runtime command — setup, install-hooks, allow (exceptions), unlock (agent self-edit window), status, doctor
- [Layered Policy & Exemptions](https://www.prismor.dev/docs/policy-layers-and-exemptions): Org/project/repo precedence, the non-overridable floor, explicit enforce selection, and the self-protection rules
- [Patchbot](https://www.prismor.dev/docs/patchbot): Open-source dependency vulnerability scanner that opens a fix PR per package (CLI, GitHub Action, or scheduled Managed Agents)

## Products

- [MCP Security](https://www.prismor.dev/mcp): Security gateway for MCP servers and AI IDEs
- [Supply Chain Security](https://www.prismor.dev/supply-chain): Dependency scanning, SBOM analysis, and vulnerability auto-fix
- [CRA Compliance](https://www.prismor.dev/cra): EU Cyber Resilience Act compliance automation
- [SBOM & VEX](https://www.prismor.dev/cra-sbom): Software Bill of Materials generation with SPDX 2.3
- [Pricing](https://www.prismor.dev/pricing): Plans and pricing

## Blog

- [Security Blog](https://www.prismor.dev/blog): AI agent security, runtime threats, and supply chain research

## Company

- [Homepage](https://www.prismor.dev): Main website
- [Privacy Policy](https://www.prismor.dev/privacy): Data protection and privacy practices
- [GitHub App](https://github.com/apps/prismor-auto-security-scan-and-fix): Install the Prismor GitHub App

## Contact

- Website: https://www.prismor.dev
- Twitter: https://x.com/prismor_dev
- Email: contact@prismor.dev