Features

One page for each screen and feature of DevHub: what it does, when to use it and how.

• Dashboard: the MCP server, the last 24 hours of calls and what needs attention.
• MCP server: start, stop and restart it, copy its address, choose the tools.
• Organizations and projects: group your work and choose the folders agents may open.
• Vaults: notes for an organization or a repository, and what agents may do with them.
• Vault Viewer: read, edit, search and link notes, graph, canvas, PDF.
• Business Context: search decisions and rules across your vaults.
• AI Context: queue what an agent should know and check its size.

• Repository Index: symbols and relations of your code, callers, impact and routes.
• Library: skills, commands, personas, rules, user profiles and the library graph.
• Instructions: procedures you hand to an agent by name.
• Analyses: documents, variants and findings from looking into something before building it.
• Plans: steps with gates and a file scope, and the check against git.
• Agents: one role, one agent; run it and watch its sessions.
• Agent Intelligence: memory, instincts and what reaches an agent.
• Activity: the local log of every tool call, and what it never stores.

• Harnesses: connect AI clients and set up a folder for each one.
• Vartari Hub (experimental): choose which model does which job.
• Settings: the eight sections of Settings.
• Accessibility: theme, font, text size, zoom, contrast and motion.
• Desktop Pet: a character that reacts to agent activity.
• Updates: checking, downloading and installing a new version.
• Plans and tiers: Standard, Plus and Premium, and permissions.

For the ideas behind these screens, see Concepts.