A: Claude Code on macOS: yes. It’s one of the agents vexp auto-configures, and it’s one of the most mature integrations we have, so it works well on macOS. You install vexp, run setup, and it wires up Claude Code for you.

ShadowGit: it’s complementary, not a competitor, because it solves a different problem. ShadowGit gives your AI the time dimension of your code: it auto-captures every change (a parallel git history) and lets your agent query what changed, what worked before, and restore any past version. It’s about history and debugging. vexp gives your AI the structural dimension: how your code connects right now (what calls what, what depends on what), and feeds the agent only the relevant slice for a task while cutting tokens.

So one answers “what changed and when,” the other answers “what depends on what and what’s needed for this task.” They run well together: both local, both over MCP, both attach to Claude Code. Your agent can use ShadowGit to understand history and vexp to understand structure in the same session. No conflict, they cover different axes.