Cyberse Framework an AI assisted development architecture III

Hello again, in this session I will continue the development of the Cyberse Framework while I explain concepts from the framework and guidelines, in previous sessions already was covered:

Cyberse introduction: here

Cyberse initial version of the cli tool using the framework: here

What elements are in an Office?

same as real work, when a team of professionals are working in a project, in first post series we illustrate it with Martin (ai developer) in songs-inc-office:

songs-inc-office/
└── team
    └── martin
        ├── body
        ├── CLAUDE.md -> head/identity.md
        ├── GEMINI.md -> head/identity.md
        └── head
            ├── identity.md
            ├── long-memory.md
            └── short-memory.md

5 directories, 5 files

and in second post series we illustrate the office concept with Ray (AI Node-js developer) in cyberse-office:

cyberse-office
└── team
    └── ray
        ├── body
        │   ├── index.js
        │   └── package.json
        ├── CLAUDE.md -> head/identity.md
        ├── GEMINI.md -> head/identity.md
        └── head
            ├── identity.md
            ├── long-memory.md
            └── short-memory.md

5 directories, 7 files

but so far we only has been talking about team element inside the office, a team is the central element in a office because as in real life an office without a team is just an empty space, but an office has more qualities beyond a team, lets review some of them:

• An office serve a business purpose.
• An office is a facility for a team to successfully achieve their work.
• An office has multiple elements that serves above's purpose.
• Some elements are: desks, white-board, meeting-room, office library, archive room, etc.

Whiteboard

In Cyberse framework those elements exists and pursue the same purpose, lets talk about whiteboard, a white board serve to share information relevant to everybody and make it visible for them, at same time that information is ephemeral and not long time persistent, like immediate goals, lets create and write down into white board for Cyberse office:

I have created the whiteboard.md file in cyberse-office:

  Today is May 14, 2026, team goals:                                          
                                                                              
  [x] Writing CLI tool to create a new Cyberse project                        
  [ ] CLI tool should create empty files short-memory.md long-memory.md,      
  identity.md.                                                                
  [ ] Create office white-board.md                                            
  [ ] e2e test that validates above's files created succesfully via cli                                  
                                                                              
  Let's do it team!                                                          

cyberse-office folder tree:

.
├── team
│   └── ray
│       ├── body
│       │   ├── index.js
│       │   ├── package.json
│       │   ├── package-lock.json
│       │   └── test
│       │       └── cyberse.spec.js
│       ├── GEMINI.md -> head/identity.md
│       └── head
│           ├── identity.md
│           ├── long-memory.md
│           └── short-memory.md
└── whiteboard.md

6 directories, 10 files

Above white-board.md serves to set a goals and communicate important imformation relevant for the team, like the date.

Office bookshelf

lets talk about bookshelf element in the office, this element is where all important books are stored, those books are related to daily operations that any team member can access, update, add, remove, different from each team member's memory and identity, this bookshelf contains assets relevant for everyone in the team, for instance Architectural Decision Records or ADR, so lets create the bookshelf with the empty space to store our future ADR's:

.
├── bookshelf
│   └── adr
│       ├── index.md
│       └── template.md
├── team
│   └── ray
│       ├── body
│       │   ├── index.js
│       │   ├── package.json
│       │   ├── package-lock.json
│       │   └── test
│       │       └── cyberse.spec.js
│       ├── CLAUDE.md -> head/identity.md
│       ├── GEMINI.md -> head/identity.md
│       └── head
│           ├── identity.md
│           ├── long-memory.md
│           └── short-memory.md
└── whiteboard.md

8 directories, 12 files

so our bookshelf has an adr folder where there is an empty index (no ADRs right now) and a template.md, the template looks like:

  # {Short title - ADR 0000}                                                  
                                                                              
  ## Status                                                                   
                                                                              
  accepted | proposed | rejected | deprecated | superseded by ADR-0123        
                                                                              
  ## Considered options                                                       
                                                                              
  • option 1                                                                  
  • option 2                                                                  
  • ...                                                                       
                                                                              
  ## Decision                                                                 
                                                                              
  Choosen {option X}, because {justification}                                 
                                                                              
  ### Pros                                                                    
                                                                              
  if any                                                                      
                                                                              
  ### Cons                                                                    
                                                                              
  if any                                                                      
                                                                              
  ## More information                                                         

Summary

in this short session we analyzed two new office's elements:

• white-board
• bookshelf

both of them serves the same purpose, share and store short and large chunks of context that agents can check, update, add, remove, over the time, than elements creates the context necessary for the AI agents work in their daily tasks successfully but without necessity of load everything at once, those elements can be accessed on demand depending of the task, so each agent will store in their memory what is, where, and how to use them.

In a next session I will continue expanding more those concepts while working with agents to create outcomes like the cyberse-cli tool.