Hexis

Product Introduction

1. Definition: Hexis is an open-source, Git-backed platform for centralized AI skill, tool, and knowledge management. Technically, it is a Model Context Protocol (MCP) server and governance layer that provides version-controlled, access-controlled resources to AI agents.
2. Core Value Proposition: Hexis exists to solve enterprise AI agent sprawl and governance. It provides a single source of truth for company-approved AI capabilities, enabling secure, scalable, and vendor-agnostic AI agent deployments with full audit trails and change control.

Main Features

1. Git-Centric Knowledge & Skill Base: All company knowledge, AI skills (defined in SKILL.md files), and tool manuals (UTCP) are stored in a dedicated Git repository owned by the company. This enables native version history, branching, and pull request/change request workflows for all modifications, applying software development best practices to AI agent management.
2. Centralized MCP Server with OAuth 2.1: Hexis runs a built-in remote MCP server. Employees connect their AI agents (Claude Code, Cursor, ChatGPT, etc.) once using a single connection key. The server dynamically serves only the skills, tools, and context the user's role permits, eliminating per-tool credential management and per-agent setup.
3. Role-Based Access Control (RBAC) & Governance: Administrators define granular access policies in a roles.yaml file. Changes to protected branches require approval via change requests, ensuring all modifications to AI capabilities are reviewed. This governance model applies uniformly to both human users and AI agents.

Problems Solved

1. Pain Point: Fragmented and insecure AI agent tooling. Companies face chaos from employees manually sharing API keys, duplicating skill prompts, and using unvetted tools across different AI agents, leading to security risks, inconsistency, and loss of institutional knowledge.
2. Target Audience: AI Engineering Teams, DevOps/Platform Engineers, Enterprise IT/Infosec teams, and business unit leaders (e.g., Sales, Marketing, Engineering) rolling out AI agents to their teams. Personas include the "AI Platform Engineer" who needs to govern tools, the "Sales Manager" who wants to equip their team with approved sales skills, and the "Developer" who consumes these skills daily.
3. Use Cases: 1) Onboarding a new team to company AI agents in minutes instead of weeks. 2) Rolling out a new, vetted data analysis tool to the finance team's agents securely. 3) An AI agent suggesting an improvement to a skill, which is automatically submitted as a pull request for owner review. 4) Switching AI agent vendors (e.g., from ChatGPT to Claude) without rebuilding the toolchain.

Unique Advantages

1. Differentiation: Unlike siloed vendor ecosystems (e.g., OpenAI's GPTs) or manual prompt libraries, Hexis is built on open protocols (MCP, Git) and is self-hostable. It decouples AI capability management from any specific agent vendor, future-proofing the investment and preventing lock-in.
2. Key Innovation: The integration of enterprise-grade Git workflows (branching, PRs, approval gates) with the AI agent tooling layer (MCP). This treats AI skills and knowledge as first-class, versioned code assets, enabling the governance, collaboration, and auditability that enterprises require but which native agent platforms lack.

Frequently Asked Questions (FAQ)

1. What is Hexis and how does it work with AI agents like Claude? Hexis is a central management platform that hosts your company's AI skills and tools. It works by acting as a standardized MCP server that agents like Claude Desktop or Cursor can connect to. Once connected, the agent can automatically discover and use the approved skills and knowledge bases you've configured in Hexis, without any further setup.
2. Is Hexis a SaaS product or self-hosted software? Hexis is open-source software (Apache 2.0 license) designed primarily for self-hosting via Docker. The company Bevel also offers a managed SaaS version. The self-hosted deployment gives you full control over your data, Git repository, and infrastructure.
3. How does Hexis handle security and access control for AI tools? Hexis provides role-based access control (RBAC) managed through Git. Tool credentials are stored in an encrypted vault within the platform. Access to specific tools and knowledge areas is defined per role, and all changes, including those proposed by AI agents, flow through a pull request review process before being merged and activated.
4. Can I use Hexis with my existing GitHub or GitLab repository? Yes. Hexis is designed to work with any standard Git host (GitHub, GitLab, Bitbucket, Azure DevOps, or self-hosted). You provide the HTTPS clone URL and a personal access token with read/write permissions during setup, and Hexis will manage the repository's content.
5. What happens if I switch from one AI agent (e.g., ChatGPT) to another (e.g., Claude)? Since Hexis uses the open Model Context Protocol (MCP), your investment in skills and tools is preserved. You simply connect the new agent to your Hexis MCP server endpoint, and it immediately has access to the same governed set of capabilities, enabling a vendor-agnostic AI strategy.