AGENT-GLOBAL-BASE

A Markdown-first, agent-native knowledge base — an organizational simulation and LLM engineering laboratory running under one governance framework. There is no root build system, no entry point, and no startup script. The primary artifacts are documents: agent profiles, skill specifications, pipeline definitions, and Python reference implementations.

---

The Three Co-Resident Systems

System	Path	What it is
The Company	company/	Mobile product company — departments, pipelines, crew
The Studio	studio/	Casual Games Studio — 39 crew, 11-stage game pipeline
CC-00 Lab	core-component-00/	Applied LLM research lab — five-module engineering stack

These systems are architecturally independent but share governance through the Agent Systems Engineering (ASE) framework.

---

Repository Map

AGENT-GLOBAL-BASE/
├── AGENTS.md                    ← Full orientation guide for AI agents (read this first)
├── CLAUDE.md                    ← Claude Code operating rules (global)
│
├── company/                     ← The Company
│   ├── library/                 ← Central knowledge hub — start here for company context
│   ├── departments/             ← Agent profiles + skill files (canonical source)
│   ├── pipeline/                ← Mobile / Web / Backend API / Full-Stack / Recruitment
│   ├── recruitment/             ← Hiring cycles and templates
│   ├── optimization-history/    ← Append-only archive of optimization plans
│   └── project/                 ← Active project dashboard
│
├── studio/
│   └── casual-games/            ← Only active studio
│       ├── library/             ← Studio overview, charter, strategic brief
│       ├── pipeline/            ← 11-stage game development pipeline
│       ├── team/crew/           ← 38 FTE + 1 Contract across 7 divisions
│       └── projects/            ← Per-game folders (none scaffolded yet)
│
├── core-component-00/           ← CC-00 Lab (only place with runnable code)
│   ├── agent-systems-engineering/  ← ASE governing framework
│   ├── engineering/                ← Layers 1, 2, 3, 5
│   │   ├── prompt-engineering/         ← Layer 1 — what to write
│   │   ├── context-engineering/        ← Layer 2 — how to structure it
│   │   ├── harness-engineering/        ← Layer 3 — how to execute safely
│   │   └── multi-agent-engineering/    ← Layer 5 — how agents cooperate
│   ├── retrieval-augmented-generation/  ← Layer 4 — where to get content
│   ├── mcp-servers/                ← MCP server implementations (deployment surface)
│   └── crew/                       ← Lab Director + 11 crew (director/elias-vance/, module leads, etc.)
│
└── telescope/                   ← Cross-department research index (company/, core-component-00/,
                                     and studio/casual-games/ each keep their own instance too —
                                     see telescope/README.md)

---

The Company

Start here: company/library/README.md

Departments & C-Suite

Department	C-Suite Owner	Core Focus
Brand Design	CDO — Yuki Tanaka-Chen	Mobile UI/UX, prototyping, IDS
Cyberspace Security	CIO — Dr. Priya Mehta / CSO — Dr. Sarah Chen	Security architecture, risk
Human Resources	CHRO — Dr. Evelyn Hartwell	Recruitment, onboarding
Legal	CLO — Dr. Victoria Svensson-Park	Governance, contracts, privacy
Localization	CTO-L — Dr. Amara Osei-Mensah	i18n pipeline, TMS, translation
Product Management	CPO — Marcus Tran-Yoshida	PRD authorship, product strategy
Research & Development	CTO — Dr. Kenji Nakamura	Android, iOS, KMP, Flutter, testing

13-Stage Development Pipeline

#	Stage	Key Owner	Gate
0	Problem Validation	CPO	—
1	Requirements → PRD + SRD	CPO, CSO	✅
2	PRD → Prototype + IDS	CDO	✅
3	Prototype → UML Engineering Pkg	CTO, CIO	✅
4	UML → Implementation Plan + Gantt	CTO	✅
5	Plan → Software Development	CTO	—
6	Development → Arch. Review	CTO + full panel	✅
7	Arch. Review → Automated Testing	CTO + Test Lead	✅
8	Testing → Integrity Verification	CTO + all C-suite	—
9	Integrity → Translation Production	CTO-L + R&D	—
9.5	Internal Dogfood	VP Quality	—
10	Translation → Release Readiness	CTO + User	✅
11	Live Operations (continuous)	VP Platform	QBR

Four pipeline variants exist — Mobile, Web, Backend API, Full-Stack — all built on a shared _base/ skeleton with variant-specific deltas. Recruitment is a separate standalone 9-stage pipeline.

---

The Casual Games Studio

Start here: studio/casual-games/library/overview/casual-games-studio.md

• Engine: Unity 6.3 LTS
• Crew: 39 people across 7 divisions (Leadership · Production · Creative-Design · Art · Audio · Engineering · Live-Ops)
• Pipeline: 11 stages (Stage 0–10) — distinct from the company's 13-stage pipeline
• Projects: No games scaffolded yet; project folders live under studio/casual-games/projects/

Studio Director: Dr. Marcus Vogel · Creative Director: Sakura Ishimori

---

Core Component 00 — LLM Engineering Lab

Start here: core-component-00/README.md

CC-00 is the organisation's centralised LLM engineering laboratory. Every team building with large language models starts here. Lab Director: Dr. Elias Vance.

The Five-Module Stack

Layer	Module	Has Runnable Code
1	engineering/prompt-engineering/	No
2	engineering/context-engineering/	Yes
3	engineering/harness-engineering/	Yes
4	retrieval-augmented-generation/	Yes
5	engineering/multi-agent-engineering/	Yes

Key production implementations live under each module's implementations/ folder. The governing framework — ASE — is defined in core-component-00/agent-systems-engineering/.

---

Key Conventions

Item	Convention
Folder names	kebab-case — always
Agent profile	agent/profile.md — six required YAML frontmatter fields
Skill files	skills/<skill-name>.md — adjacent to the agent folder
Pipeline document	pipeline.md — inside the pipeline folder; the canonical source
Optimization records	company/optimization-history/YYYY-MM-DD-<slug>/ — append-only
Markdown formatting	Run prettier --write "<file>" on every created or modified file

Document authority when sources conflict: pipeline.md → agent/profile.md → library/overview/*.md → library/departments/*.md

---

Working with Organizational Agents

Agent personas (CDO, CTO, Studio Director, Lab Director, etc.) are documents, not running processes. To get output from a named agent:

1. Read their agent/profile.md — establishes identity and authority scope.
2. Read all referenced skills/*.md — these are executable contracts, not suggestions.
3. Adopt their voice and produce output strictly within their documented authority.

Profile locations:

• Company C-suite → company/departments/<dept>/supervisor/<role>/agent/profile.md
• Studio crew → studio/casual-games/team/crew/<division>/<role>/<name>/agent/profile.md
• Lab Director → core-component-00/crew/director/elias-vance/agent/profile.md

---

Quick Navigation

I need…	Go to
Full workspace orientation (agents)	AGENTS.md
Company overview / people / pipeline	company/library/README.md
A specific department's agents	company/departments/
Studio structure + game pipeline	studio/casual-games/library/overview/casual-games-studio.md
LLM engineering patterns	core-component-00/README.md
ASE governance (ADRs, compliance)	core-component-00/agent-systems-engineering/
Production Python code	core-component-00/engineering/<module>/implementations/
MCP server implementations	core-component-00/mcp-servers/
Research archive (cross-department index)	telescope/README.md
CC-00 engineering + LLM research	core-component-00/telescope/README.md
Company product research	company/telescope/README.md
Studio game/market research	studio/casual-games/telescope/README.md