
Open-source git-backed control plane for enterprise AI agents — context, skills, tools and permissions as files in your own repo, served over MCP.
Open-source git-backed control plane for enterprise AI agents — context, skills, tools and permissions as files in your own repo, served over MCP.
Hexis is the open-source core of Bevel, a vendor-agnostic control plane for enterprise AI agents. Instead of scattering context, skills, tools, and permissions across each vendor's console (Claude, Cursor, ChatGPT, opencode), Hexis makes them files in your own git repository and serves them to any agent runtime over MCP (or UTCP, the open protocol Bevel's team authored). Knowledge lives as typed nodes with provenance, procedures live as plain Markdown skills, tool manifests declare secrets and access rules once, and each agent is a named identity with its own credentials — no shared service accounts. Changes go through review like any other code. Bevel (the paid platform) adds hosted services, integrations, and enterprise support, while Hexis stays free and open-source for teams who want to self-host the reusable context layer.
Hexis is an open-source, Git-backed control plane designed for enterprise AI agents, providing context, skills, tools, and permissions organized as files in a personal repository, which are then served over a Multi-Cloud Platform (MCP). This structure enhances collaboration and efficiency in AI development.
Hexis serves as a centralized control plane for managing enterprise AI agents, making it easier to configure their behavior and permissions. By leveraging Git, Hexis allows teams to maintain version control over their AI agent configurations, which is crucial for tracking changes, reverting to previous states, and collaborative development.
Context Management: Hexis allows you to define the context in which AI agents operate. This means developers can tailor the environment and rules under which an AI agent performs tasks, leading to improved performance and relevance.
Skills and Tools: Organizations can define specific skills and tools that AI agents can utilize, enhancing their capabilities. For example, an AI agent designed for customer support can be equipped with natural language processing tools and access to a knowledge base.
Permissions Control: The platform allows for granular permissions, ensuring that only authorized users can modify settings or access sensitive data, thereby enhancing security in enterprise environments.
Hexis works by integrating a Git-backed source of truth that maintains context, skills, tools, and permissions as Markdown/YAML files in your repository. This allows enterprises to manage AI agents across vendors with a structured knowledge layer, ensuring accountability and reducing vendor lock-in.
Hexis operates by utilizing a Git-backed repository where essential components such as context, skills, tools, and permissions are stored as Markdown or YAML files. This structure allows enterprises to manage their AI environments more effectively, as it provides version control similar to traditional code.
Markdown/YAML Files: All context and skills are stored in easily readable formats. This makes it simple for domain owners to review and update procedures without sifting through complex vendor configurations.
Typed Knowledge with Provenance: Every fact in Hexis is represented as a typed node that includes its source, owner, and last verification date. This transparency enables teams to track the integrity and timeline of their data, facilitating mass updates and comprehensive dashboards for analysis.
Tool Manifests with Vaulted Secrets: Tools are declared once within the system, with sensitive information securely stored in a vault. File-level access rules ensure that only authorized agents can access specific files and endpoints, enhancing security and control over actions taken by AI agents.
Per-Agent Identity: Each agent operates under its own identity, eliminating the risks associated with shared service accounts. This means every action taken by an agent is traceable back to a unique set of credentials, ensuring accountability and compliance.
Hexis offers several key features, including a Git-backed source of truth, typed knowledge with provenance, and Markdown skills formatted as readable documents. Additionally, it provides tool manifests with vaulted secrets and distinct per-agent identities, ensuring secure and organized management of knowledge and tools in AI-driven environments.
Hexis is designed to streamline the management of AI tools and agents, ensuring that all components are easily accessible and maintainable.
Git-Backed Source of Truth: This feature allows users to treat their knowledge base like code. By storing context, skills, tools, permissions, and agent identities as Markdown or YAML files in a Git repository, users can review changes, create branches, and conduct diffs just like with regular code, promoting collaboration and version control.
Typed Knowledge with Provenance: Each piece of knowledge in Hexis is represented as a typed node in a graph structure. This means that every fact is not only organized but also carries metadata such as its source, owner, and last verification date. This ensures that users can easily trace the origin of information, facilitating better decision-making and accountability.
Markdown Skills: Unlike traditional AI systems that rely on prompt fragments hidden in vendor configurations, Hexis allows domain experts to document procedures as clear, readable Markdown files. This transparency makes it easier for team members to review and update processes, ultimately enhancing operational efficiency.
Tool Manifests with Vaulted Secrets: With Hexis, tools are declared once in a manifest, while secrets are securely stored in a vault. This setup allows fine-grained access control, where file-level rules determine which agents can read specific files and invoke certain endpoints.
Per-Agent Identity: Each agent in Hexis operates with a unique identity and set of credentials. This eliminates the use of shared service accounts, ensuring that every action is directly attributable to a specific agent, which enhances security and accountability.
Hexis is designed for enterprises utilizing AI agents across various vendors like Claude, Cursor, and ChatGPT. It benefits teams involved in procurement, engineering, and platform management by providing a shared knowledge layer, preventing vendor lock-in, and enabling better control over AI agent identities and access.
Hexis serves a diverse range of users who need optimized management of AI agents across different platforms. Organizations can effectively manage and deploy AI agents from multiple vendors, such as Claude, Cursor, and ChatGPT, leveraging a unified infrastructure.
Enterprises Running AI Agents: Businesses utilizing AI agents from various vendors can benefit significantly from Hexis. It allows them to define context and skills just once. This eliminates the need to re-upload these elements for each vendor, streamlining operations and enhancing efficiency.
Teams Building Specialized Agents: Procurement, Go-To-Market (GTM), and Request for Information (RFI) teams can create specialized agents, such as tender desks and campaign agents, on a shared knowledge layer. This layer maintains provenance, ensuring that all agents are grounded in a consistent knowledge base, which can be reused across different projects.
Control and Accountability: Platform teams can assign per-agent identities and file-level access controls. This means that every tool call made by an AI agent is traceable, enhancing accountability. Furthermore, destructive endpoints are protected by a review process, ensuring that any significant changes undergo proper scrutiny.
Vendor Lock-In Prevention: Hexis provides a portable specification (MCP/UTCP) that enables companies to switch between different agent runtimes without losing their defined context and skills. This feature is crucial for organizations that prioritize flexibility and adaptability in their AI strategies.
Transparency and Review: Engineering leaders can utilize Hexis to maintain an "AI operating manual" that is subject to review through Git diffs and change requests. This transparency is vital for continuous improvement and for ensuring that the AI operates according to predefined standards.
Hexis offers a free tier that allows users to access basic features, while its paid plans start at competitive pricing for advanced functionalities. The exact cost varies depending on the plan selected, with options tailored to different user needs.
Hexis is known for its flexible pricing model that caters to a diverse audience. The free tier enables users to explore essential features without any financial commitment, making it ideal for individuals or small teams just starting with the platform.
For users needing more robust capabilities, Hexis offers several paid plans. Typically, these plans range from $10 to $100 per month, depending on the level of service and features required. For instance:
Hexis also frequently updates its offerings, so it’s advisable to check their website for the latest pricing and features.
To get started with Hexis, visit Bevel Software to sign up for an account. Once registered, you can explore Hexis' features, tools, and resources designed for effective project management and collaboration.
Hexis is a powerful platform for project management that streamlines collaboration among teams. To get started, follow these steps:
Compare Hexis: vs TryCase · vs Aside · vs Wisry · vs Cline