Command Center · the ops console
The console I run everything from
The console is the bridge; Claude Code is the engine room.
A local-first operations console where the UI and the AI work the same files. It's where projects, tasks, agent assignments, tools, and architectures live — the source that several sections of this very site are exported from.
Local-first
No database — the truth is plain JSON files, one per record. Anything that can read files can read (and audit) the whole operation.
Live
A file watcher pushes changes to the browser over SSE. The screen moves the instant either the UI or an agent touches anything.
Self-briefing
A nightly headless agent sweeps every source and writes the morning briefing before I am awake — current without being tended.
Human-gated
Agents dispatch, claim, and finish work — but only I move it to done. That gate is enforced in code, not convention.
What it holds
projects — the ventures and builds in motion tasks — the kanban of what moves each forward tools — the registry behind the Tools page architectures — the diagrams behind the Architectures page assignments — work dispatched to agents (inbox → active → review → done) goals — now / this-year / multi-year learning — what I am actively getting fluent in knowledge — an index into the vault briefings — the nightly self-written summaries brain — a sanitized graph of the knowledge vault How it ties in here
This public site never connects to the command center live — it's gated and private by design. Instead, a reviewed public-safe snapshot is exported from it: the 9 tools and 3 architectures you can browse here were published through an opt-in gate that strips anything sensitive before it ever reaches the web.