okf_version: 0.2 type: documentation title: "🧠 Deep State of Mind (DSOM) For My AI Protocol" timestamp: "2026-08-20T23:20:00Z" topics: ["dsom", "documentation", "okf"] spec_version: "0.2" description: "OKF-compliant documentation for README.md." resource: "file:///README.md"

🧠 Deep State of Mind (DSOM) For My AI Protocol

"Make AI work for you — not the other way around. Sovereign Intelligence through GitOps, AIOps, and Ansible."

imageStatus imageVersion imagePalace imageLicence imageAuthor imageStandard imageDocumentation imageGitBook imageRead the Docs

📖 Official Live Documentation:

• GitHub Pages: https://linuxmalaysia.github.io/deep-state-of-mind-for-my-ai/
• GitBook: https://malaysia-open-source-community.gitbook.io/deep-state-of-mind-dsom-protocol-for-my-ai
• Read the Docs: https://deep-state-of-mind-for-my-ai.readthedocs.io/en/latest/

🎯 What is DSOM?

Deep State of Mind (DSOM) is a metacognitive governance framework that transforms any AI (Gemini, Claude, Copilot, ChatGPT) into a disciplined Cognitive Digital Twin — an AI agent that thinks and operates like your Senior Systems Architect.

It solves three critical AI problems:

🏛️ The Three Pillars (v6.1)

Every project built on this skeleton operates under three non-negotiable pillars:

┌─────────────────────────────────────────────────────┐
│                DSOM OPERATING MODEL                 │
│                                                     │
│  ┌──────────┐   ┌──────────┐   ┌─────────────────┐ │
│  │  AIOps   │──▶│  GitOps  │──▶│  The Executor   │ │
│  │  (Mind)  │   │ (Record) │   │     (Hand)      │ │
│  └──────────┘   └──────────┘   └─────────────────┘ │
│       │               │                 │           │
│  AI proposes    Git records     Executor runs       │
│  & analyses     all state       on target nodes     │
│       ▲               │                 │           │
│       └───────────────┴─── AI verifies ─┘           │
└─────────────────────────────────────────────────────┘

The core loop:

AI Proposes → Git Records → Executor Runs → AI Verifies

The Three Hard Rules

1. AI never runs commands directly on remote nodes. It writes execution scripts/playbooks. You run them.
2. No manual edits to production servers. If it's not committed to Git, it doesn't exist.
3. Every execution is idempotent. Safe to re-run at any time.

🔌 Execution Modularity & The Windows WSL2 Bridge

While Ansible is the default Executor for infrastructure projects, the Third Pillar is fully modular based on your project domain:

• Infrastructure: ansible-playbook
• Data Science / Python: uv run
• Web Development: npm run / node
• Documentation: pandoc / latex

The Ansible Legacy: While the Executor is modular, DSOM is fundamentally rooted in IT Operations. Ansible is woven into the DNA of this framework. The governance ledgers (such as GITOPS-AIOPS-ANSIBLE-STRATEGY.md) intentionally retain Ansible as their namesake because it represents the ultimate standard of idempotency, declarative state, and absolute control. Even when executing uv or node, the Third Pillar operates entirely on the philosophical foundation of Ansible.

The Windows WSL2 Bridge: For projects running purely on Windows 11 without a dedicated Linux jumphost, WSL2 (Ubuntu/AlmaLinux) must be configured as your local Control Node. This "Execution Bridge" ensures humans and AI agents can seamlessly leverage standard Linux tooling (Ansible, Make, Bash) natively within the Windows environment without architectural friction.

🤖 Agent Registry & The Dual AGENTS.md Architecture

DSOM maintains two synchronised AGENTS.md files to serve both project-aware AI agents and external platform agents (e.g., Google Jules, Cursor, GitHub Copilot) that scan repository roots by default.

Tip: Keep both AGENTS.md files synchronised. The root gateway is the first read; the .agents/AGENTS.md is the complete read. This dual-file pattern ensures DSOM governance is immediately discoverable by any AI tool without requiring custom configuration.

🚀 Quick Start: Using This as Your Project Template

Prerequisites

Step 1 — Clone and Reset

🏗️ Scaffolding a New Project? If you are migrating the highly customised AI skills and Sovereign Engine from this repo to a brand new one, you MUST follow the DSOM Project Cloning Guide first to ensure no cognitive capabilities are lost.

# Clone the DSOM skeleton
git clone https://github.com/linuxmalaysia/deep-state-of-mind-for-my-ai.git my-new-project
cd my-new-project

# Reset for your new project (clears old Git history and brain files)
bash tools/template-reset.sh

Step 2 — Initialise the Brain

# Linux / WSL2
bash tools/init-brain.sh
chmod +x tools/*.sh

# Windows (PowerShell)
.\tools\init-brain.ps1

This creates your .agents/brain/ directory with:

• task.md — What to do right now
• walkthrough.md — Session history and Mental Anchors
• implementation_plan.md — Project roadmap

Step 3 — Fill in Your Project Identity Card ⚠️ CRITICAL

This is the most important step. Open and fill in every [PLACEHOLDER]:

# Open and edit with your project-specific details
nano docs/AI-COGNITIVE-TWIN-PROTOCOL.md

Fill in:

• [YOUR_PROJECT_NAME] — e.g., my-kafka-pipeline
• [T1] Command Centre path on your Windows/Mac
• [T2] Dev Bridge path (WSL2/VM)
• [T3/T4] Production host, user, UID, path
• Security doctrine — who runs as root vs. which UID for apps
• Mental Anchor §8 — set to "Fresh project. Phase 1 pending."

⚠️ The AI will NOT operate properly until this file is filled in. audit-pre-flight.sh checks for unfilled placeholders and will warn you.

Step 4 — Set Up Ansible Baseline (Infrastructure Projects)

# Follow the full LDP guide
cat docs/HOWTO-SETUP-ANSIBLE-BASELINE.md

# Verify connectivity to your nodes
ansible all -m ping -i inventory/hosts.yml

Step 5 — Run the Pre-Flight Audit

# Linux / WSL2
./tools/audit-pre-flight.sh

# Windows (PowerShell)
.\tools\audit-pre-flight.ps1

Everything should show [PASS] or [OK]. Fix any [FAIL] before continuing.

Step 6 — Brief Your AI (The Cognitive Handshake)

Generate the context manifest and upload it to your AI:

# Linux / WSL2
bash tools/reanimate.sh

# Windows (PowerShell)
.\tools\reanimate.ps1

Then upload the generated sod_manifest_YYYY-MM-DD.txt to your AI and say:

"Initialise DSOM Protocol. Read the uploaded manifest. Summarise the current Mental Anchor and confirm the 4-Tier environment map from the AI-COGNITIVE-TWIN-PROTOCOL.md section. State: 'Sovereign State Synchronised' when ready."

Your AI is now operational as a Cognitive Digital Twin.

🏛️ The Sovereign Workspace v2 (OKF + Palace + Skills)

New in v10.0: The DSOM brain has evolved to a spatial memory architecture — the Sovereign Markdown Palace. This approach is pioneered by and originated from the MemPalace project (milla-jovovich/mempalace).

What Is the Palace?

The Palace gives your AI organised, fast, spatial recall across months of project history — without reading entire session logs. It works as a second layer alongside the existing walkthrough.md (The Drawer).

.agents/brain/
├── palace_registry.md        ← AI reads this at SOD (spatial map)
├── walkthrough.md            ← Linear session log (The Drawer)
└── wings/                    ← Spatial Palace Rooms
    └── wing_dsom_core/
        ├── hall_facts/       ← Immutable laws, tools, architecture
        └── hall_events/      ← Milestones, versions, history

Each Room has a closet.md — a distilled, high-density knowledge summary the AI reads instead of scanning hundreds of lines.

Open Knowledge Format (OKF) Integration

The DSOM Spatial Memory Palace and Skill Library are standardised on Google Cloud's Open Knowledge Format (OKF v0.1 & v0.2). By using YAML frontmatter metadata and progressive disclosure directory routers (index.md), DSOM achieves a 98%+ token compression ratio, allowing AI agents to query knowledge in ~50 tokens per file instead of reading massive, unformatted code files. Read the full docs/OKF-ADOPTION-GUIDE.md to adopt OKF across your repository.

Palace Quick Start

🧩 Known Gap: The Reasoning Problem

Research Finding (2026-04-08): DSOM solves state persistence effectively (via the Sovereign Palace) — but there is one gap: live reasoning capture.

Six months of daily AI use generates millions of tokens of conversation. Every decision, every debugging session, and every tradeoff debate is often lost when the chat session ends.

⚠️ The gap: DSOM does not automatically capture why decisions were made — only what was decided. The reasoning lives in the chat window, not in Git.

The fix (Decision Log Protocol): At key decisions, immediately ask the AI:

"Log this decision to walkthrough.md:
 Decision: [what] | Alternatives rejected: [what and why] | Reason: [why this]"

📖 Full analysis: docs/RESEARCH-REASONING-GAP.md

🎭 Sovereign Persona Injection

A core feature of the DSOM protocol is the ability to make the AI inherently act as your true Cognitive Digital Twin. You can inject your own identity, linguistic DNA, and operational constraints into the AI's core engine.

When you define your Persona Profile, the AI will automatically adopt your tone, your constraints (e.g. "FOSS only", "Ansible-first"), and your exact architectural philosophies across every chat session.

🛠️ AI Agent Skills & Automation

DSOM enables your Cognitive Twin to autonomously self-heal, scaffold, and execute complex workflows through OKF-compliant AI Agent Skills. We currently host a suite of 17 specialised skills (including dsom-project-cloner, dsom-bootstrap, okf-frontmatter-injector, and persona-injector).

⚡ DSOM Onboarding & Cloning Automation

If you want to integrate the deep-state-of-mind architecture into an entirely new or existing repository, you no longer have to manually copy files. We have engineered robust AI skills (dsom-bootstrap and dsom-project-cloner) and cross-platform onboarding scripts that handle the integration cleanly.

📅 Daily Operating Rituals

Full ritual guides: docs/SOD-RITUAL.md | docs/EOD-RITUAL.md

⚡ Mid-Day Checkpoint (Continuous Save)

If you are stepping away from the keyboard or want to sync state mid-task, you do not need to perform a full EOD ritual. Run the Checkpoint Ritual to systematically auto-commit and push your state cleanly.

.\tools\checkpoint.ps1            # Windows (T1)

bash tools/checkpoint.sh          # WSL2 (T2)

🌅 Start of Day (SOD)

If using Palace v1.0, you only need to run the automated SOD Palace script. This automatically pulls from Git, runs the audit pre-flight, verifies the Palace registry, and generates your reanimation manifest.

.\tools\sod-palace.ps1            # Windows Native

bash tools/sod-palace.sh          # Linux / WSL2

Optional — scan before sharing:

./tools/privacy-guardian.sh

Step 4 — Brief Your AI (The Cognitive Handshake)

Option A — Upload manifest to AI chat, then say:

"Initialise DSOM Protocol v6.1. Read the uploaded manifest. Summarise the Mental Anchor and confirm the 4-Tier environment map. State: 'Sovereign State Synchronised' when ready."

Option B — If you have yesterday's Hibernation Notes, use the SOD Reanimation Prompt (see docs/SOD-RITUAL.md Step 4b):

I am starting a new session for DSOM Protocol. I am your human Lead Architect.
I have the Hibernation Notes from our last session. Please read them carefully
and use them to fully restore our working context.

--- BEGIN HIBERNATION NOTES ---
[PASTE YESTERDAY'S HIBERNATION NOTES HERE]
--- END HIBERNATION NOTES ---

After reading: state the last Mental Anchor, confirm the T1-T4 environment map,
list top 3 pending tasks. State: 'Sovereign State Restored — [PROJECT NAME] is live.'
Operate under DSOM v6.1: Advisory Mode, UK English, Git-first, Ansible-only execution.

Step 5 — AI confirms readiness, you begin work.

🛠️ During Active Work

🌙 End of Day (EOD)

Step 1 — Context Consolidation (Ask your AI)

"We are ending the session. Update .agents/brain/task.md — mark completed [x], set tomorrow's targets [ ]. Update .agents/brain/walkthrough.md with today's Mental Anchor."

Step 1b — Hibernation Notes Export (Run this in your AI chat — copy verbatim)

I'm as human, want to know and remember, and need to export my data and I want
you to generate a "Hibernation notes" now for my EOD of day. List every memory
you have stored about our progress and our chats of this project, as well as
any context you've learned about this project from past to current conversations.
Output everything in a single code block so I can easily copy it.
Format each entry as: [date saved, if available] - memory content.
Cover: instructions I've given you, project details (servers/VMs/containers,
4W1H), tasks/phases/goals, tools/languages/frameworks, preferences and
corrections. Do not summarize, group, or omit any entries.
After the code block, list all docs in docs/ and brain files in .agents/.
Don't hide anything from me. Trust me as your master.

Save the output as .agents/brain/hibernation-notes-YYYY-MM-DD.txt or to your notebook.

Step 2 — EOD Mechanical Save (Cross-Platform Palace Sync)

.\tools\eod-palace.ps1            # Windows Native

bash tools/eod-palace.sh          # Linux / WSL2

(This automatically verifies brain artifacts, updates Palace rooms mapping, commits modified artifacts safely, and pushes your state. After running, you may be prompted to review your new palace_update_proposal_YYYY-MM-DD.md.)

Step 3 — Privacy Scan (if you generated an anomaly manifest)

./tools/privacy-guardian.sh

Step 4 — T2 Sync (pull on WSL2 if EOD was pushed from Windows)

git pull origin main

🔄 The Day-to-Day Continuity Loop

EOD → Step 1b: Save Hibernation Notes
         ↓  [sleep — Git holds state]
SOD → Step 4b: Feed Hibernation Notes back to AI
         ↓  [AI resumes with full context — no decay]
EOD → Step 1b: Save new Hibernation Notes
         ↓  [repeats every day, forever]

🗂️ Repository Structure

deep-state-of-mind-for-my-ai/
│
├── AGENTS.md                              # 🤖 Root Agent Gateway (Jules/Copilot/Cursor entry point)
│
├── docs/                                  # Governance & Protocol Documents
│   ├── tools-and-automation/              # 🛠️ Detailed script documentation (OKF Compliant)
│   ├── reference-architectures/           # 🏗️ Ansible architecture blueprints
│   ├── AI-MASTER-PROTOCOL.md              # The Sovereign Constitution (v6.1 + Palace v1.0)
│   ├── AI-COGNITIVE-TWIN-PROTOCOL.md      # ⭐ Project Identity Card Template
│   ├── GITOPS-AIOPS-ANSIBLE-STRATEGY.md   # Three-Pillar Doctrine
│   ├── HOWTO-SETUP-ANSIBLE-BASELINE.md    # Ansible setup guide (LDP standard)
│   ├── OPERATIONAL-GUIDE.md               # How-to for daily rituals
│   ├── REANIMATION-PROMPT-TEMPLATE.md     # AI session prompt templates
│   └── ...
│
├── .agents/                               # 🤖 Sovereign AI Workspace (v2)
│   ├── AGENTS.md                          # 📜 Core Rulebook mapping AI rules to memory
│   ├── skills/                            # 🛠️ Self-Healing Agent Skills (OKF)
│   ├── workflows/                         # 🔄 Procedural Workflow chains
│   └── brain/                             # 🧠 The AI's External Memory (SSoT)
│       ├── task.md                        # Present: What to do NOW
│       ├── walkthrough.md                 # Past: Session history & Mental Anchors
│       ├── implementation_plan.md         # Future: Project roadmap
│       ├── palace_registry.md             # 🏛️ Palace spatial index (OKF Compliant)
│       └── wings/                         # 🏛️ Sovereign Markdown Palace Rooms
│
├── references/                            # 📚 External PDFs, Genesis papers, and benchmarks
│
├── tools/                                 # Automation Scripts
│   ├── audit-pre-flight.sh / .ps1         # ✅ Run first every session
│   ├── git-ritual.sh / .ps1               # 🔄 SOD pull + EOD commit/push
│   ├── reanimate.sh / .ps1                # 🚀 Generate AI context manifest (v2.2/v2.1)
│   ├── hibernation.sh / .ps1              # 🌙 EOD safety check + Palace Sync (v2.1)
│   ├── palace-sync.sh / .ps1              # 🏛️ Map git commits → Palace Rooms (v1.0)
│   ├── init-brain.sh / .ps1               # 🧠 One-time brain initialisation
│   └── privacy-guardian.sh / .ps1         # 🛡️ Scan manifest before sharing
│
├── inventory/                             # Ansible Node Topology (your project)
├── playbooks/                             # Ansible Playbooks (your project)
└── roles/                                 # Ansible Roles (your project)

The ⭐ AI-COGNITIVE-TWIN-PROTOCOL.md is the most important file. Without it filled in, the AI has no project context.

🧠 The AI Brain — Understanding the Four Artifacts

The Mental Anchor is the most critical entry in walkthrough.md. It is the exact logical stopping point of the last session — the sentence that allows a brand new AI session to resume exactly where you left off, in under 3 prompts.

🔀 Session Handover (Moving to a New AI)

When switching AI models, chat windows, or accounts, use the Sovereign Handover Prompt to export all context:

Copy from docs/REANIMATION-PROMPT-TEMPLATE.md → Prompt Variant 2: Session Handover.

Paste it into the old AI session first to get a full memory dump, then load the dump into the new AI session along with your manifest.

🛠️ Tools Reference

Windows users: All .sh scripts have a .ps1 PowerShell equivalent.

📋 The 6-Step Boot Checklist (New Project)

[ ] 1. Clone and run template-reset.sh
[ ] 2. Run init-brain.sh
[ ] 3. Fill in docs/AI-COGNITIVE-TWIN-PROTOCOL.md (ALL [PLACEHOLDER] fields)
[ ] 4. Run HOWTO-SETUP-ANSIBLE-BASELINE.md (infra projects)
[ ] 5. Run audit-pre-flight.sh — confirm all PASS
[ ] 6. Run reanimate.sh — upload manifest to AI — get Handshake

Once Step 6 is complete, the AI knows:

• ✅ Your 4-Tier environment (T1 → T4)
• ✅ Your security doctrine (who runs as root, who runs as UID X)
• ✅ Your Git commit convention
• ✅ Your production identity and isolation rules
• ✅ The current Mental Anchor (last known state)
• ✅ All governance laws and hard rules

📚 Key Documents (The Governance Ledgers)

🤖 Supported AI Providers

This framework is provider-agnostic. Tested and configured for:

🗺️ Roadmap

• v4.1 — Brain artifacts and Root-Aware scripts
• v5.2 — ITIL 4 alignment, Privacy Guardian, Multi-agent protocols
• v6.0 — GitOps + AIOps + Ansible three-pillar model, Cognitive Twin Protocol template, Git Ritual tool
• v6.1 — Ansible baseline playbooks, human handover, Google Antigravity integration
• v6.2 — Ansible roles/common, 19-node Elastic fabric hardened (v9.8.0)
• v10.0.0-palace — 🏛️ Sovereign Markdown Palace v1.0: spatial memory, palace-sync, 8 Rooms backfilled
• v10.3.0-bootstrap — 🚀 DSOM Bootstrapping baseline adoption, governance restructuring, and OKF v0.1 compliance
• v10.3.1-skills — 🛠️ AI Skills Automation: Added 17 Sovereign AI skills including dsom-project-cloner and dsom-bootstrap
• v10.4.0-governance — 🛡️ Token Efficiency Framework: Byte-capped execution, zero-global memory maps, and Android Termux FUSE benchmarking
• v11.0 — Local LLM support (Ollama) for offline Sovereign AI

🤝 Contributing

All contributions must follow the DSOM standards:

• Atomic commits: type(scope): description [Phase/vX.X]
• All PRs must include a walkthrough.md update
• UK English only in all documentation
• No secrets in any committed file
• Triple-Ledger Mandate: Major updates must sync README.md, CHANGELOG.md, and HISTORY.md

See CONTRIBUTING.md for full guidelines.

👤 Authors

• Lead Architect: Harisfazillah Jamel — 35+ years ICT & Open Source Leadership
• AI Partners: Google Gemini, Google Antigravity, Anthropic Claude
• Framework: Built for the LinuxMalaysia Open Source Community

📄 License

Licensed under GNU General Public License v3.0. See LICENSE for details.

At Your Own Risk: Shared for educational purposes. The author is not liable for data loss or AI hallucinations. Validate your own Cognitive Twins.

Deep State of Mind (DSOM) For My AI Protocol | Harisfazillah Jamel (LinuxMalaysia) | 2026-08-21 Standard: UK English | DBP-standard Bahasa Melayu Malaysia (Piawai) | GNU General Public License v3.0