CommandHive allows role based access to AI agents, and connect them to MCP servers.
We are building role based AI agents, companies, universities have different identities and credentials say for an university, a student and teacher. Now everyone in the university should not have same level of access to an AI agent.
As AI agents become more personalised and fine tuned there is a need to restrict the AI agent responses based on credentials and identity of a person prompting it. We do not want children to have tools for nefarious activities but at the same time, a teacher should have maximum access possible. The agent being finetuned on responses of teacher and student, needs to access the same model. (relevant context)
CommandHive solves exactly this.
The entire project was built during the hackathon, we had already created the pip package https://pypi.org/project/bee-agent/ which is A2A orchestration engine but it's not being used, we also created the home page previously. https://commandhive.xyz/
We are completely bootstrapped