🕯️ DSOM Ritual of Transition (v6.1 + Palace v1.0)
docs/RITUAL-OF-TRANSITION.md
"Intelligence is ephemeral; the Repository is eternal." Standard: DSOM For My AI Protocol v6.1 + Palace v1.0 | GitOps · AIOps · Ansible
The Ritual of Transition governs movement between cognitive states (AI context) and physical states (Git, Ansible, infrastructure). It is the operating membrane of the three-pillar model.
🏛️ 1. The Three-Pillar Transition Loop
Every state transition in DSOM passes through this loop. There are no shortcuts.
┌─────────────────────────────────────────────────────────┐
│ DSOM TRANSITION LOOP │
│ │
│ [SOD: Reanimate] → [Propose] → [Git Record] → │
│ [Ansible Execute] → [Verify] → [EOD: Hibernate] │
│ ↑_______________________________________↓ │
│ (repeats every day, forever) │
└─────────────────────────────────────────────────────────┘
| Phase | Pillar | Action |
|---|---|---|
| 🌅 SOD Reanimation | AIOps | Load context → Cognitive Handshake |
| 🛠️ Active Flow | AIOps + GitOps | AI proposes → Human approves → Git records |
| ⚙️ Execution | Ansible | Human runs Ansible → AI verifies output |
| 🌙 EOD Hibernation | GitOps | Sovereign Save → Hibernation Notes → git push |
🌅 2. Phase 1: Reanimation (Start-of-Day Transition)
Objective: Restore the Cognitive Twin's full context from Git and initiate the Sovereign Directive.
Sequence:
OPTION A: The Ansible Shortcut (Recommended on T2)
Step 1: bash tools/sod-palace.sh # (Windows: .\tools\sod-palace.ps1)
→ Pulls Git, audits workspace, generates manifest with Section [14] Palace Registry
Step 2: Upload manifest to AI → Cognitive Handshake
OPTION B: Manual / Windows T1
Step 1: git-ritual.sh sod → pull latest from origin/main
Step 2: audit-pre-flight.sh → verify workspace (5 checks, all must PASS)
Step 3: reanimate.sh → generate 14-section context manifest
Step 4: privacy-guardian.sh → scan manifest before upload
Step 5: Upload manifest to AI → Cognitive Handshake
Stop Conditions — Reanimation FAILS if:
- AI cannot state the correct Mental Anchor from
walkthrough.md - AI references wrong environment (wrong T-tier, wrong OS, wrong user)
audit-pre-flight.shshows any[FAIL]— fix first, reanimate after- AI proposes to execute directly on production nodes without Ansible
The Sovereign Directive (confirm readiness):
"Sovereign State Synchronised — [PROJECT NAME] is live and ready."
If the AI cannot say this correctly, the SOD is incomplete. Re-run from Step 3.
Full SOD guide: docs/SOD-RITUAL.md
🛠️ 3. Phase 2: Active Flow (Execution Guardrails)
Objective: Maintain the three-pillar model and architectural integrity during active work.
3.1 The Human-AI Division of Responsibility
| Human (Sovereign Architect) | AI (Cognitive Digital Twin) |
|---|---|
| Defines the mission | Proposes implementation plan |
| Approves all proposals | Explains Why before What |
| Commits to Git | Updates .agents/brain/task.md |
| Runs Ansible commands | Writes the exact command for human |
| Pastes output back to AI | Verifies output, proposes next step |
3.2 The Active Flow Rules
- One thing at a time — AI proposes a single step. Human approves. Execute. Then next step.
- Git before execution — every change is committed before
ansible-playbookruns. - Brain before code — update
.agents/brain/task.mdto reflect what you're doing before starting. - Output before proceeding — AI does not assume success. Human pastes terminal output every time.
- Why before What — AI always explains the architectural reason before showing the command.
3.3 Ansible-First Rule (OS-Level Changes)
| ✅ Correct | ❌ Wrong |
|---|---|
ansible-playbook playbooks/dsom/[name].yml -i inventory/hosts.yml |
ssh root@node "systemctl restart service" |
ansible-lint playbooks/dsom/[name].yml before commit |
Manual config edits directly on nodes |
--check dry-run before real run |
Ad-hoc ansible -m shell -a 'cmd' on production |
3.4 Stop Conditions During Active Flow
Immediately pause and re-read docs/AI-MASTER-PROTOCOL.md if:
- AI suggests running commands directly on production nodes
- AI suggests committing secrets or tokens to any file
- AI skips explaining the architectural reason for a change
- AI proposes a change that would affect more than one logical concern (not atomic)
- Human is about to
git push --forcetomain
🌙 4. Phase 3: Hibernation (End-of-Day Transition)
Objective: Crystallise the day's work into Git and preserve the Cognitive Twin context for tomorrow.
Sequence:
Step 1: AI updates task.md → mark [x] done, [ ] tomorrow's targets
Step 1b: AI updates walkthrough.md → new Session Anchor with Mental Anchor sentence
Step 2: Run EOD Hibernation Notes prompt → save output to hibernation-notes-YYYY-MM-DD.txt
OPTION A: The Ansible EOD (Recommended on T2)
Step 3: bash tools/eod-palace.sh # (Windows: .\tools\eod-palace.ps1)
→ Validates artifacts, runs palace-sync, semantic commit, and git push automatically
OPTION B: Manual / Windows T1
Step 3: hibernation.sh → safety check (dirty state, uncommitted files) + palace-sync.pl
Step 4: privacy-guardian.sh → scan if manifest was generated today
Step 5: git-ritual.sh → interactive EOD commit (guided semantic commit + push)
Step 6: git pull origin main → T2 WSL2 sync after T1 push
Step 7: Write tomorrow's Mental Anchor in your physical/digital notebook
The Sovereign Save Commit format:
chore(brain): EOD sovereign save — [Short Anchor] [Phase/vX.X]
Example:
chore(brain): EOD sovereign save — AlmaLinux 10 WSL setup complete [Phase-1/v6.1.1]
Stop Conditions — EOD INCOMPLETE if:
task.mdhas no[x]items for todaywalkthrough.mdis missing a new Session Anchorhibernation.shreturns any non-GREEN statusgit pushhas not completed successfully- T2 WSL2 has not pulled from T1 push
Do not sleep until all stop conditions are resolved.
Full EOD guide: docs/EOD-RITUAL.md
🔀 5. The AI Model Transition (Switching AI Providers)
Objective: Transfer the Cognitive Twin state from one AI model to another without context loss.
This is triggered when:
- The current AI model is degrading (context decay, repeated errors)
- Switching between AI providers (Gemini → Claude → Copilot → Antigravity)
- Starting a fresh conversation in the same session
The Transition Sequence:
Step 1: In the outgoing AI session — run EOD Hibernation Notes Export (Prompt 5)
Step 2: Copy the full Hibernation Notes code block
Step 3: Open the new AI session
Step 4: Paste Prompt 3 (Cognitive Twin Handover) + Hibernation Notes block
Step 5: AI confirms Mental Anchor, T-tier map, sovereign laws, top 3 tasks
Step 6: AI states: "Cognitive Twin Transfer Complete — [PROJECT NAME] Handover Successful."
Why this works: The Hibernation Notes block contains everything the AI needs — project identity, environment, history, pending tasks, and operating rules. No manifest upload needed. No file access needed. Just text.
Full prompt: docs/REANIMATION-PROMPT-TEMPLATE.md → Prompt 3
🏛️ 6. The Sunday Human Refresh Protocol
Every Sunday, the Lead Architect performs a Dry-Run Audit — a full human re-read of the project state:
# Run the full DSOM audit — confirm all 5 checks pass
./tools/audit-pre-flight.sh
# Or via Ansible (checks all inventory nodes simultaneously)
ansible-playbook playbooks/dsom/audit-preflight.yml -i localhost,
# Review the last 10 commits
git log --oneline -10
# Read the current Mental Anchor
cat .agents/brain/walkthrough.md | tail -30
The Human Re-Index Checklist:
[ ] AI-MASTER-PROTOCOL.md — governance laws still apply
[ ] AI-COGNITIVE-TWIN-PROTOCOL.md — all [PLACEHOLDER] fields filled
[ ] task.md — accurate, no stale items
[ ] walkthrough.md — Mental Anchor is current and precise
[ ] implementation_plan.md — roadmap still reflects reality
[ ] audit-pre-flight.sh — all 5 checks [PASS]
[ ] CONTRIBUTING.md — new contributors can follow it
🔗 7. Related Documents (Read in This Order)
| Priority | Document | When |
|---|---|---|
| 1 | docs/AI-MASTER-PROTOCOL.md |
Every session |
| 2 | docs/SOD-RITUAL.md |
Every morning |
| 3 | docs/EOD-RITUAL.md |
Every evening |
| 4 | docs/REANIMATION-PROMPT-TEMPLATE.md |
When prompting AI |
| 5 | docs/HUMAN-HANDOVER-CONTEXT.md |
New session, new AI |
| 6 | docs/AI-COGNITIVE-TWIN-PROTOCOL.md |
Per project — fill once |
Standard: DSOM For My AI Protocol v6.1 + Palace v1.0 | Harisfazillah Jamel | LinuxMalaysia This is the baseline transition protocol for all projects built on this skeleton. Last Updated: 2026-04-08 | Version: v6.1
Deep State of Mind (DSOM) For My AI Protocol | Harisfazillah Jamel (LinuxMalaysia) | 2026-07-04 Standard: UK English | DBP-standard Bahasa Melayu Malaysia (Piawai) | GNU General Public License v3.0