Article ID: ChronoFlight.RuntimePromptPack.v1_0.CF
Version: v1.0
Status: Canonical / Almost-Code / Prompt Runtime Spec
Scale: Civilisation / Dual
Domain: Prompting / Human–AI Coordination / Command Templates / Runtime Execution
Mode: Mixed / CFCS-ready
ChronoFlight Lens: Structure × Phase × Time

AI Ingestion Lock

This article does not create a new primitive.

It defines the canonical prompt layer for running the ChronoFlight Control Tower through human prompts and LLM command forms.

It sits directly on top of the already-locked runtime stack:

ChronoFlight Control Tower Runtime v1.0
Kernel Sensor Pack
Kernel Trigger Matrix
Kernel Coupling Atlas
Runtime Glossary v1.0
the 13-lane kernel spine
the shared state grammar
the fixed alert words
the fixed action verbs
the fixed priority trees

This page does one thing:

It freezes the correct way to ask the runtime to act.

Classical Foundation Block

A runtime weakens when people know the framework but prompt it badly.

A system can have:

good sensors
good trigger logic
good control grammar

and still fail in practice if the instructions given to the operator or LLM are:

vague
overly broad
semantically unstable
missing the trigger
missing the time slice
missing the desired output shape

So a real runtime needs not only:

control terms
control sheets
scenario runs

It also needs:

stable prompt forms

That is the classical foundation of this article.

Civilisation-Grade Definition

The ChronoFlight Runtime Prompt Pack v1.0 is the canonical set of human prompts and LLM command templates used to run the Control Tower layer, so triggers, chains, priority trees, cut-points, and repair routes can be requested in one stable, low-ambiguity format.

In simple form:

same runtime
same language
same prompt shape
repeatable output

That is the core definition.

CORE CLAIM

A civilisation-grade control runtime becomes materially more usable when humans and LLMs share a small, frozen set of prompt shapes that reliably produce the right kind of runtime output under load.

That is the main lock.

PURPOSE OF THE PROMPT PACK

This prompt pack exists to do six things:

reduce prompting ambiguity
preserve runtime semantics
shorten human-to-AI command time
standardize output shape
prevent essay-style drift when action is needed
make the Control Tower layer runnable by non-experts once the grammar is learned

This is not a general prompting guide.

It is the runtime prompt layer.

PROMPT DESIGN RULE (LOCKED)

A valid runtime prompt should contain, in the smallest viable form:

time slice
trigger or objective
scope
desired output mode

Core rule

Prompt the runtime for state, triage, and action—not for broad freeform discussion—when operating inside the control layer.

That is the design rule.

CANONICAL PROMPT BLOCK (LOCKED)

Every strong runtime prompt should be reducible to this base structure:

PROMPT CORE

Time Slice:
Scope:
Trigger or Task:
Output Mode:

This is the minimum prompt skeleton.

CANONICAL OUTPUT MODES (LOCKED)

The prompt pack assumes six standard output modes.

1. STATE READ

Return:

trigger
active chain
floor status
current lane states

2. TRIAGE READ

Return:

priority tree
anchor lanes
best cut-point

3. FIRST-RESPONSE PACK

Return:

protect
cut
fence
route
hold
escalate-if
success condition

4. VALIDATION CHECK

Return:

is the response working?
is the cut working?
is escalation required?
is widening allowed?

5. SCENARIO RUN

Return:

a worked simulated runtime cycle

6. COMPARATIVE READ

Return:

compare two entities, slices, or route states using the same control grammar

These modes are fixed.

PROMPT CLASS SYSTEM (LOCKED)

The prompt pack is divided into seven classes:

Lane Check Prompts
Full Kernel Prompts
Trigger Prompts
Repair Routing Prompts
Validation Prompts
Scenario / Simulation Prompts
Compression / Dashboard Prompts

These seven classes are sufficient for Control Tower use.

1. LANE CHECK PROMPTS (LOCKED)

Use these when checking one lane only.

Prompt Form A1 — Single Lane State Check

Template

Time Slice: . Check . Return lane state, alert band, route state, main drift, main repair, and whether it is propagating hazard.

Example

Time Slice: current week. Check WaterOS. Return lane state, alert band, route state, main drift, main repair, and whether it is propagating hazard.

Use

local diagnostics
early warning
lane-specific monitoring

Prompt Form A2 — Single Lane Trigger Check

Template

Time Slice: . is at . Return anchor, cut-point, and first-response pack only.

Example

Time Slice: current slice. EnergyOS is at Fence. Return anchor, cut-point, and first-response pack only.

Use

fast triage
high-pressure lane response

Prompt Form A3 — Single Lane Repair Check

Template

Time Slice: . For , assume current action is . Return whether the lane is stabilizing, still descending, or needs escalation.

Example

Time Slice: next slice. For LogisticsOS, assume current action is “Cut low-value movement and hold critical route only.” Return whether the lane is stabilizing, still descending, or needs escalation.

Use

next-slice validation
hold / escalate decisions

2. FULL KERNEL PROMPTS (LOCKED)

Use these when reading the whole spine.

Prompt Form B1 — Full Kernel Slice Read

Template

Time Slice: . Read the full 13-lane kernel. Return primary trigger, active failure chain, strongest coupling, floor status, and priority tree.

Example

Time Slice: current month. Read the full 13-lane kernel. Return primary trigger, active failure chain, strongest coupling, floor status, and priority tree.

Use

system-wide diagnostics
control-tower start state

Prompt Form B2 — Full Kernel Action Read

Template

Time Slice: . Read the full 13-lane kernel and emit the Control Tower Sheet: trigger, chain, floors, anchors, cut-point, first-response pack, escalation condition, success condition.

Use

main runtime execution
one-sheet operating output

Prompt Form B3 — Floor Integrity Check

Template

Time Slice: . Check Survival Floor, Enabling Floor, and Control Floor. Return which floor is most endangered, why, and what must be protected first.

Use

fast triage orientation
preventing wrong tree selection

3. TRIGGER PROMPTS (LOCKED)

Use these when a known trigger has already fired.

Prompt Form C1 — Trigger-to-Action Prompt

Template

Time Slice: . Trigger: . Return only: priority tree, anchor, cut, fence, route, hold, escalate-if, success condition.

Example

Time Slice: current slice. Trigger: SecurityOS Emergency. Return only: priority tree, anchor, cut, fence, route, hold, escalate-if, success condition.

Use

direct control response
high-speed action output

Prompt Form C2 — Co-Trigger Prompt

Template

Time Slice: . Co-trigger: + at . Return which priority tree overrides, what anchor spine applies, and what must be cut first.

Example

Time Slice: current slice. Co-trigger: Standards&MeasurementOS + GovernanceOS at Fence. Return which priority tree overrides, what anchor spine applies, and what must be cut first.

Use

coupled crises
escalation logic

Prompt Form C3 — Trigger Escalation Prompt

Template

Time Slice: . Current trigger is . Assume first-response pack already applied. Return the exact condition that should force escalation in the next slice.

Use

advance preparation
avoid slow escalation

4. REPAIR ROUTING PROMPTS (LOCKED)

Use these when the trigger is known and the main question is routing.

Prompt Form D1 — Best Repair Route Prompt

Template

Time Slice: . Trigger: . Return the shortest high-leverage repair route A → B → C and explain why this route should run before broader restoration.

Use

route sequencing
narrow corridor restoration

Prompt Form D2 — Cut-Point Selection Prompt

Template

Time Slice: . Trigger: . Return the best cut-point, why it is upstream, and what propagation it reduces first.

Use

leverage selection
avoiding symptom-chasing

Prompt Form D3 — Anchor Spine Prompt

Template

Time Slice: . Trigger: . Return the smallest viable anchor spine that must not be lost before repair widens.

Use

narrowing under stress
anti-overexpansion

5. VALIDATION PROMPTS (LOCKED)

Use these after action is already underway.

Prompt Form E1 — Next-Slice Validation Prompt

Template

Time Slice: next slice. Current action pack: . Return: holding, stabilizing, failed cut, failed route, or escalate.

Use

outcome validation
action review

Prompt Form E2 — Widening Check Prompt

Template

Time Slice: . Current corridor is narrowed but stable. Return whether it is widenable yet, and if not, what must hold longer first.

Use

anti-premature recovery
corridor widening discipline

Prompt Form E3 — Failed Response Prompt

Template

Time Slice: . The current first-response pack did not reduce propagation. Return the next cut-point, stronger tree if needed, and revised route.

Use

fast reroute
post-failure adjustment

6. SCENARIO / SIMULATION PROMPTS (LOCKED)

Use these for practice, testing, and AI alignment.

Prompt Form F1 — Canonical Scenario Run Prompt

Template

Run a ChronoFlight scenario for . Return: trigger lane, active chain, priority tree, anchor lanes, best cut-point, first-response pack, and success condition.

Example

Run a ChronoFlight scenario for EnergyOS reserve collapse with WaterOS still at Watch. Return: trigger lane, active chain, priority tree, anchor lanes, best cut-point, first-response pack, and success condition.

Use

rehearsal
training
simulation

Prompt Form F2 — Branch Stress Test Prompt

Template

Simulate a two-slice ChronoFlight run where crosses Fence and becomes a co-trigger in the next slice. Show the shift in priority tree and action pack.

Use

escalation training
co-trigger logic testing

Prompt Form F3 — Scenario Comparison Prompt

Template

Compare two scenario runs: vs . Return what changes in anchor choice, cut-point, and route order.

Use

operator learning
understanding why different triggers need different first moves

7. COMPRESSION / DASHBOARD PROMPTS (LOCKED)

Use these when the output must be short and control-ready.

Prompt Form G1 — One-Line Runtime Prompt

Template

. Give short action block.

Example

WaterOS Emergency. Give short action block.

Expected Output

Short runtime answer only:

Trigger
Tree
Protect
Cut
Route
Success

Use

rapid control calls
dashboards
mobile / compressed interfaces

Prompt Form G2 — Control Tower Sheet Prompt

Template

Return the Control Tower Sheet for this slice only. No essay.

Use

direct operator dashboard output
repeated cycle execution

Prompt Form G3 — Glossary-Locked Prompt

Template

Use Runtime Glossary v1.0 terms only. Return short commands only.

Use

forcing semantic discipline
AI alignment under stress

CANONICAL HUMAN → AI PROMPT FORMS (LOCKED)

These are the shortest approved handoff forms from a human operator.

Form H1 — Direct Trigger Call

Trigger: . Give first-response pack.

Example:
Trigger: EnergyOS Fence. Give first-response pack.

Form H2 — Full Runtime Read

Read current slice. Give Control Tower Sheet.

Form H3 — Co-Trigger Call

Co-trigger: + at . Give tree, anchor, cut, route.

Form H4 — Validation Call

Next slice check. Current pack applied. Is it holding, stabilizing, or escalating?

Form H5 — Widening Call

Current corridor is stable but narrow. Is widening allowed yet?

These are the most useful operational handoff prompts.

CANONICAL AI → HUMAN ANSWER FORMS (LOCKED)

The preferred reply shape is:

SHORT CONTROL ANSWER

Trigger:
Chain:
Tree:
Protect:
Cut:
Fence:
Route:
Hold:
Escalate If:
Success:

This answer shape should remain stable across prompt classes whenever concise output is requested.

PROMPT MODIFIERS (LOCKED)

These are the only approved short modifiers for changing output style without changing runtime meaning.

SHORT

Return compressed operator answer only.

FULL

Return full Control Tower Sheet.

STRICT

Use Runtime Glossary terms only.

NO-ESSAY

Do not explain beyond action-relevant lines.

COMPARE

Return differential read between two slices / runs / entities.

SIMULATE

Treat the prompt as a scenario, not a live operational read.

These modifiers help keep prompts small and stable.

CANONICAL “DO NOT PROMPT LIKE THIS” LIST (LOCKED)

Avoid prompts like:

“What do you think is happening?”
“Can you analyze everything?”
“What should we do generally?”
“Give advice.”
“Fix the whole system.”
“Tell me what matters.”
“Monitor this.”
“Be careful.”

These are too broad or semantically weak for runtime control.

Replacement rule

Replace broad prompts with:

a trigger
a scope
an output mode

That is the anti-vagueness rule.

PROMPT ESCALATION RULE (LOCKED)

If the first prompt returns something too broad, the operator should narrow the next prompt, not broaden it.

Correct escalation sequence

broad slice read
identify trigger
ask for trigger-to-action pack
ask for validation next slice
ask for widening only after stabilization

Wrong escalation sequence

broad slice read
system feels unclear
ask for “everything in more detail”
create more narrative but less control

Core rule

When under pressure, prompts should become narrower and more action-bound, not longer and more exploratory.

CHRONOHELMAI PROMPT RULE (LOCKED)

ChronoHelmAI should be prompted in one of two modes.

Mode 1 — Read Mode

Used to identify:

trigger
chain
tree
cut-point

Mode 2 — Action Mode

Used to emit:

protect
cut
fence
route
hold
escalate-if
success condition

Core rule

Do not mix exploratory narrative and high-pressure action mode in the same prompt unless explicitly needed.

This preserves cleaner runtime execution.

FENCEOS / ERCO PROMPT FORMS (LOCKED)

FenceOS Prompt Form

Trigger: . Return hard boundaries only: what to cut, what to fence, and what must not be crossed.

Use when:

immediate threshold defense is the main need

ERCO Prompt Form

Trigger: . Assume containment is still possible. Return the smallest viable restitch route and what to hold while restitching.

Use when:

the main need is localized repair and narrow stabilization

These prompt forms keep the two layers distinct.

PROMPT PACK FOR TRAINING (LOCKED)

This pack should also be used for operator training.

The preferred training progression is:

Stage 1

Use:

single-lane prompts
short trigger calls

Stage 2

Use:

co-trigger prompts
priority tree selection prompts

Stage 3

Use:

full Control Tower Sheet prompts
next-slice validation prompts

Stage 4

Use:

scenario simulation prompts
comparison prompts
widening check prompts

This creates a clean learning curve from simple to full-runtime use.

WHAT THIS PROMPT PACK PREVENTS

This article prevents five common runtime prompting failures:

1. Prompt Vagueness

The runtime is asked something too broad to return a usable control answer.

2. Semantic Drift

Different prompt wording causes the same trigger to be handled differently.

3. Narrative Inflation

The system returns essays when short action commands are needed.

4. Human–AI Handoff Friction

Humans and AI use different prompt structures for the same control task.

5. Non-Repeatable Execution

The same situation gets different prompt shapes every time, weakening training and reliability.

This is why the prompt pack is a required runtime layer.

CANONICAL CHECKLIST

A valid use of the Runtime Prompt Pack is only acceptable if it preserves:

the prompt core block
the fixed output modes
the seven prompt classes
the short human → AI forms
the short AI → human answer shape
the approved modifiers
the anti-vagueness rule
the “narrower under pressure” escalation rule
the separation between Read Mode and Action Mode

If not, the prompt layer is drifting.

CANONICAL LOCK

ChronoFlight Runtime Prompt Pack v1.0 locks the canonical human prompts and LLM command templates for running the Control Tower layer, so the 13-lane civilisation spine can be read, triaged, and acted on through one stable prompt grammar instead of ad hoc questioning.

From this point onward:

the runtime now has a frozen prompting layer
human and AI operators can invoke the same control logic with the same short templates
and the ChronoFlight branch now includes a full prompt-to-runtime execution interface for repeatable, low-ambiguity use

This is the Runtime Prompt Pack v1.0 lock.

ONE-LINE COMPRESSION

ChronoFlight Runtime Prompt Pack v1.0 freezes the exact way humans and LLMs should ask the control tower for state, triage, and action, so the runtime can be invoked quickly, consistently, and without semantic drift.

NEXT IN SEQUENCE

The strongest next article is:

ChronoFlight Runtime Starter Kit v1.0: The Smallest Install Order for Humans, Teams, and LLMs to Use the Control Tower Without Reading the Full Branch First