AI Command Center

A personal AI partnership system. Clone this repo, fill in your identity, and every AI conversation starts knowing who you are, what mode you're in, and where everything stands.

What This Does

Every new AI chat starts from zero. You re-explain yourself, your projects, your rules, your preferences. Every time. The more chats you have, the worse it gets.

This system fixes that. It gives every AI instance:

• Your identity -- who you are, how you work, what rules to follow
• Mode awareness -- which version of you is active (writing, working, personal, research)
• Cross-session state -- what happened in other chats, what's in progress, what's next
• Work tracking -- tasks, deadlines, priorities across your whole life

You open a chat. Say one word. The AI boots as your partner.

Quick Start

1. Clone this repo and rename it (e.g., my-2026-command-center)
2. Fill in identity/identity-rules.md with your information (section headings guide you)
3. Set up Azure DevOps project for work tracking
4. Copy system/templates/project-instructions.md into your Claude project instructions, fill in PATs and repo URL
5. Start a chat. The boot sequence does the rest.

How It Works

Read system/docs/command-center-v4.3.md for the V4.3 technical specification (historical reference).

Short version:

• boot-sequence.md is the universal boot protocol. Every AI reads it first (via CLAUDE.md wrapper).
• identity/boot.json defines what files to load and where personalities live.
• identity/identity-rules.md is who you are. Loaded every time.
• identity/personalities/ define your thinking modes. Different tone, different files, different rules per mode.
• identity/tracking/areas.json tracks categories handled by existing personalities (no separate thinking mode needed).
• identity/state/status.md is the cross-session bridge. Generated by the Synthesizer extension.
• system/ is shared infrastructure. Scripts, extensions, docs. Same across all users.

Repo Structure

boot-sequence.md                       Universal boot protocol
CLAUDE.md                              1-liner wrapper (reads boot-sequence.md)
identity/
  boot.json                            Boot configuration (SST)
  identity-rules.md                    Who you are (fill this in)
  state/status.md                      Cross-session state (auto-generated)
  personalities/
    ops/
      personality.json                 Personality definition
      behavior.md                      Behavioral rules
    workstream-1/
      personality.json                 Personality definition
      behavior.md                      Behavioral rules
  tracking/
    areas.json                         Tracking areas (handled by personalities)
system/
  scripts/                             az_ops.py, personality.py
  extensions/synthesizer-v2/           State synthesis Chrome extension
  skills/                              Procedural guides
  docs/                                Architecture specs
  templates/                           Project instructions template

Personality JSON Fields

Each personality is defined by a personality.json file:

Field	Description
name	Display name
description	One-line summary
area_path	Azure DevOps area path for work items
folder	Root-level content folder (null if none)
trigger_words	Words that activate this personality during boot
git_identity	Name and email for git commits
boot_files	Files loaded at every boot for this personality
boot_directories	Directories whose files are loaded at boot
resources	Keyword-mapped on-demand reference paths (not loaded at boot)
wiki_pages	Keyword-mapped wiki page names (not loaded at boot)
active	Whether this personality is active
created	Date created

Example Personalities and Tracking Areas

This repo ships with examples showing the V5.0 structure:

Personalities (thinking modes with their own chat sessions):

• ops -- default personality (your command center, handles everything that doesn't have its own mode)
• workstream-1 -- example personality (a separate thinking mode with its own behavioral rules)

Tracking Areas (tracked under a personality, no separate thinking mode):

• Workstream-2 -- example tracking area handled by Ops
• Workstream-3 -- example tracking area nested under Workstream-1

Ideas (just a file, no infrastructure):

• workstream-4 -- example idea in ideas/

To promote an idea to a tracking area, add it to identity/tracking/areas.json. To promote a tracking area to a full personality, create a folder in identity/personalities/ with personality.json and behavior.md.

Replace these with your actual personalities and tracking areas as you set up your system.

MCP Server

An MCP server that exposes this system to any MCP-compatible client is at ai-command-center-mcp.

License

MIT

About

Built by Nikhil Singhal at AI Trust Commons. The AI Command Center is the reference implementation of the Intent Layer architecture described in the HIP Charter.