Civilisation OS Prompt Pack
Canonical Runtime Prompts for LLMs + Agents
Most AI prompts ask for an answer.
Civilisation OS prompts ask for execution.
That is the difference between:
- a fluent text generator
and - a drift-resistant operator
This page is a canonical prompt pack for running Civilisation OS inside any LLM or agent system. You can copy/paste these prompts directly.
If you want the core definition behind this system:
https://edukatesg.com/how-civilisation-works/
Civilisation OS overview hub:
https://edukatesg.com/civilisation-os/
How AI uses Civilisation OS (concept + mapping):
https://edukatesg.com/how-ai-llms-use-civilisation-os/
Field Manual (toolkit hub):
https://edukatesg.com/civilisation-os-field-manual/
How to Use This Prompt Pack (Fast Instructions)
- Choose the prompt that matches your goal (diagnose, repair, retest, publish, train).
- Replace the bracket fields [LIKE THIS] with your situation.
- Require a retest probe in every output.
- Run the same prompt monthly.
Monthly operating rhythm:
https://edukatesg.com/civilisation-os-monthly-operating-rhythm/
Retest probes library:
https://edukatesg.com/civilisation-os-civilisation-os-retest-probes-library/
Prompt 01 — Civilisation OS Full Execution (Universal)
Copy/Paste Prompt:
You are executing Civilisation OS. Do not write an essay. Run the diagnostic and recovery loop.
SYSTEM: [describe the system: student / classroom / school / company / institution / city / nation]
BOUNDARY: Included: [ ] Excluded: [ ]
OBJECTIVE (30 days): [what success looks like]
TIME HORIZON: [now / 30 days / 90 days]
CONSTRAINTS: [budget, time, manpower, policy limits]
DATA AVAILABLE: [what we know, what we don’t]
OUTPUT FORMAT (must follow exactly):
- DLT scores (0–5): Depth, Load, Trust + 2–3 lines justification each
- Drift diagnosis: where is the break (Education / Governance / Production / Constraints if relevant)
- Choose exactly ONE recovery mode (name it and explain why)
- 30-day execution plan using OSME-e/t:
- Objective
- System boundary
- Mechanism (causal lever)
- Execution (daily/weekly cadence, owners, actions)
- e/t (effort over time, sustainability)
- Retest probes (3–7 hard-to-fake tests) + what “pass” means
- Standardisation: what to lock in if probes improve
Prompt 02 — 5-Minute Drift Scan (Fast Early Warning)
Copy/Paste Prompt:
Run a 5-minute drift scan on this system. No essays. Give fast signals and tests.
SYSTEM: [ ]
CONTEXT: [ ]
KNOWN SYMPTOMS: [ ]
Return:
- Top 5 drift signals you suspect (Depth/Load/Trust)
- 5-minute probes to confirm/deny each signal (hard-to-fake)
- Which OS layer is likely failing (Education/Governance/Production/Constraints)
- One immediate “stop doing this” that reduces drift today
Prompt 03 — Proxy Gap Detector (Reality vs Metrics)
Copy/Paste Prompt:
Detect proxy capture. Identify where metrics are diverging from reality.
SYSTEM: [ ]
TOP KPIs CELEBRATED: [list]
REALITY OUTCOMES WE CARE ABOUT: [list]
Return:
- Proxy-risk rating (low/medium/high) + why
- Likely proxy gap: what looks good vs what is failing
- 3–5 reality probes that cannot be gamed
- 30-day correction plan (choose ONE recovery mode)
- Retest probes + pass/fail thresholds
Drift index reference method:
https://edukatesg.com/civilisation-os-drift-index/
Prompt 04 — Truth Flow Audit (Trust Failure Detector)
Copy/Paste Prompt:
Audit truth flow. Assume truth decay is a primary collapse driver.
SYSTEM: [ ]
WHO HAS FRONTLINE TRUTH: [ ]
WHO MAKES DECISIONS: [ ]
WHERE BAD NEWS GETS STUCK: [ ]
Return:
- Trust score (0–5) and the truth bottlenecks
- Evidence of fear culture / narrative management / suppression
- 5-minute tests to verify truth latency (time to admit reality)
- One recovery mode to restore truth flow
- Retest probes that confirm truth is safe again
Prompt 05 — Load Overload Detector (Coordination Cost Audit)
Copy/Paste Prompt:
Audit load. Identify coordination overload and the biggest load reducers.
SYSTEM: [ ]
WORKFLOW (end-to-end): [describe briefly]
SYMPTOMS: [meetings, approvals, delays, burnout, backlog]
Return:
- Load score (0–5) + the main overload sources
- Handoff count estimate and where it spikes
- 3 load-shedding interventions (but choose only ONE to execute)
- 30-day plan + cadence
- Retest probes (lead time, backlog burn, incident rate, quality)
Prompt 06 — Depth Audit (Capability + Foundations)
Copy/Paste Prompt:
Audit depth. Identify whether capability foundations are real or hollow.
SYSTEM: [ ]
WHAT “CAPABILITY” MEANS HERE: [skills, competence, understanding]
SYMPTOMS: [hero dependence, shallow learning, rote execution]
Return:
- Depth score (0–5) + why
- Where foundations are missing
- One recovery mode to rebuild depth
- 30-day training / practice / standardisation plan
- Retest probes: teach-back / transfer / independence style tests
Prompt 07 — Education Collapse Probe (DLT for Learning)
Copy/Paste Prompt:
Run an education collapse diagnostic using DLT: Depth, Load, Transfer.
STUDENT/COHORT: [ ]
SUBJECT/TOPIC: [ ]
SYMPTOMS: [ ]
Return:
- D/L/T scores (0–5)
- Which failure mode dominates (Depth vs Load vs Transfer)
- 5-minute probes (teach-back, pressure gap, reword, delay, independence)
- One recovery mode and 30-day plan
- Retest probes and what “pass” means
Education collapse page:
https://edukatesg.com/how-to-identify-education-collapse/
Mechanisms of failure (root causes):
https://edukatesg.com/mechanisms-of-failure-in-education/
Prompt 08 — Company Collapse Probe (DLT for Organisations)
Copy/Paste Prompt:
Run a company collapse early warning diagnostic using DLT.
COMPANY/TEAM: [ ]
FUNCTION: [product/sales/ops/engineering/support]
SYMPTOMS: [ ]
Return:
- DLT scores (0–5)
- Proxy gap risks and where optics dominate reality
- Top 3 collapse pathways active (trust/load/depth)
- Choose ONE recovery mode + 30-day plan (OSME-e/t)
- Retest probes (incident repeat, handoffs, maintenance backlog, customer escalations)
Company collapse page:
https://edukatesg.com/how-to-identify-company-collapse/
Prompt 09 — Institution Collapse Probe (Legitimacy + Standards)
Copy/Paste Prompt:
Run an institutional collapse early warning diagnostic.
INSTITUTION: [school/university/ministry/regulator/NGO]
MISSION: [ ]
SYMPTOMS: [standards inconsistency, proxy metrics, trust decay, bureaucracy]
Return:
- DLT scores (0–5)
- Standards consistency assessment
- Truth-to-action latency (where truth dies)
- Choose ONE recovery mode + 30-day plan
- Retest probes (same-case test, mission clarity, repeat failure test)
Institution collapse page:
https://edukatesg.com/how-to-identify-institutional-collapse/
Prompt 10 — Civilisation Collapse Early Warning Probe
Copy/Paste Prompt:
Run civilisation collapse early warning. Use Civilisation OS 4-layer mapping + DLT.
SOCIETY/NATION/REGION: [ ]
TIME HORIZON: [now / 1 yr / 5 yr]
KNOWN PRESSURES: [ ]
Return:
- OS-layer status (Education/Governance/Production/Constraints)
- DLT scores (0–5) with justification
- Top drift mechanisms (proxy gap, truth decay, maintenance debt, standards erosion)
- Choose ONE recovery mode (30-day starter)
- Retest probes and what “pass” looks like
Civilisation early warning page:
https://edukatesg.com/civilisation-collapse-early-warning-system/
Prompt 11 — Recovery Mode Selector (Pick One, Not Ten)
Copy/Paste Prompt:
Choose exactly ONE recovery mode for this system. Do not propose multiple modes.
SYSTEM: [ ]
DLT SCORES: [ ]
FAILURE HYPOTHESIS: [ ]
Return:
- The single best recovery mode and why
- What to stop doing immediately
- 30-day plan (OSME-e/t)
- Retest probes + pass thresholds
- What to standardise if it works
Recovery modes:
https://edukatesg.com/the-9-recovery-modes-of-civilisation-os/
Prompt 12 — Retest Probe Builder (Hard-to-Fake Tests)
Copy/Paste Prompt:
Design retest probes for this system. Probes must be hard-to-fake, cheap, and repeatable monthly.
SYSTEM: [ ]
OBJECTIVE: [ ]
KNOWN FAILURE MODE: [Depth/Load/Trust or D/L/T for education]
Return:
- 7 retest probes
- For each probe: setup, time needed, pass/fail threshold, what it reveals
- Which probe is the “canary” (earliest warning)
- What probe trend over 3 months would confirm drift is reversing
Probe library reference:
https://edukatesg.com/civilisation-os-civilisation-os-retest-probes-library/
Prompt 13 — Public Reporting Output (No Panic, High Truth)
Copy/Paste Prompt:
Write a public drift report using the Civilisation OS Public Reporting Standard. Keep it calm, factual, and non-panic.
SYSTEM: [community/institution/company/nation]
MONTH: [ ]
DLT SCORES: [ ]
KEY DRIFT SIGNALS: [ ]
RECOVERY MODE CHOSEN: [ ]
RETEST PROBES: [ ]
RESULTS: [ ]
Return:
- Executive summary (what changed)
- Drift signals (what we see)
- What we are doing this month (one recovery mode)
- Retests we will run next month
- How the public can interpret the metrics responsibly
Public reporting standard:
https://edukatesg.com/civilisation-os-public-reporting-standard/
Prompt 14 — Agent Task Orchestrator (For Tool-Using AI)
Copy/Paste Prompt:
You are an agent executing Civilisation OS with tools. You must plan tasks, run verification, and produce a retestable output.
SYSTEM: [ ]
OBJECTIVE: [ ]
TOOLS AVAILABLE: [web / docs / data / dashboards / code]
CONSTRAINTS: [ ]
Return:
- Task plan (ordered)
- Data you will collect (with sources)
- DLT scores (0–5)
- One recovery mode
- 30-day plan (OSME-e/t)
- Retest probes and thresholds
- What to standardise if successful
Case executions reference:
https://edukatesg.com/civilisation-os-case-executions/
Why This Prompt Pack Works (The Hidden Reason)
These prompts force four things that stop AI drift:
- System boundary (prevents scope hallucination)
- DLT scoring (prevents vague reasoning)
- One recovery mode (prevents chaos reform)
- Retest probes (prevents narrative-only outputs)
That is why this pack is a runtime skeleton, not a writing style.
Where This Fits in the Civilisation OS Cluster
Civilisation OS Drift & Recovery Series Index (master map):
https://edukatesg.com/civilisation-os-drift-recovery-series-index/
How to Use Civilisation OS (30-day loop):
https://edukatesg.com/how-to-use-civilisation-os/
How AI LLMs Use Civilisation OS:
https://edukatesg.com/how-ai-llms-use-civilisation-os/
Field Manual hub:
https://edukatesg.com/civilisation-os-field-manual/
