Your workspace & documents

Command Center keeps everything that matters in one folder tree on your machine — the workspace root you choose at install time. Your documents, your agent definitions, your protocols, your mission records, your strategic notes: all of it is plain files in that one tree. Understanding the tree is understanding where your system actually lives.

Files are the source of truth

There is no hidden database of record. When you (or an agent) change what the system does, you change a file: a directive, a protocol, an agent definition, a routine specification. The databases that do exist — search indexes, job queues, the board projections — are caches built from the files, and they can be rebuilt from the files at any time.

This buys you three things:

• Readability. Every piece of system state is a Markdown or config file you can open in any editor.
• Shared truth. The app renders the same files the agents read and you edit — one truth, three consumers.
• Cheap change. Most behavior edits are hot: an agent or routine definition takes effect the next time one starts, with nothing to rebuild.

The shape of the workspace

Three areas of the workspace root do most of the work:

The operating contract (CLAUDE.md) sits at the root of the tree. Every agent reads it at start; it is the single most leveraged file in your workspace.

The documents tree

documents/ is where daily work accumulates. The subfolders you will touch most:

• grand-blackboard/ — your standing strategic files: active directives (what matters now), strategic context (the why), and the missions log (the mission registry). Every agent reads the directives at session start.
• general_reports/ — the landing zone for agent-produced reports, research, plans, and audits, plus the per-mission working folders under missions/.
• context-artifacts/ — the curated context packages you attach to agent sessions (see Context artifacts).
• outside_sources/ — files you bring in from outside: PDFs, data, reference material.

Lifecycle conventions — how documents flow

Agent-written reports follow a naming convention that carries the document's state in its filename: a topic, a date, a two-letter author code, and a lifecycle flag — wip (still being written), live (finished and actively referenced), done (finished, destination to be decided), stale (outdated), or archive (filed away).

New reports land in the general_reports/ root, flagged. From there, completed documents move to live/ if they stay in active use, planning/ if they are plans, or archive/ once they stop being referenced. A dedicated organizer agent handles the moves, so the landing zone stays temporary by design.

You do not have to memorize any of this to benefit from it. The Documents view surfaces the flags, and the payoff is what you experience: a workspace where the current truth is easy to find, history is preserved rather than deleted, and any agent can locate what it needs without asking you.

What this means for you day to day

• Edit freely. The files are yours. Changing a directive, a protocol, or an agent definition is the intended way to steer the system.
• Everything is portable. Your workspace is a folder. Copy it, version it, move it — the platform state travels with the files. The product does not yet include a built-in backup feature, so treat the workspace folder like any other directory you value.
• Let the conventions work. Write intent into the blackboard, let reports land where agents put them, and let the lifecycle carry documents to their long-term homes.

Where to go next

• Memory layers — what the system remembers beyond the documents themselves
• The knowledge graph (Cortex) — how the documents tree becomes a queryable graph
• Missions, operations & the fleet — the mission folders inside documents/
• Configuration reference — where settings live and what changing them costs