The Coherence Receipt Tool
An incoherence detector for structural integrity in language
This piece introduces the Coherence Receipt, a portable tool designed to be used with a language model. Drop it in, paste a text passage, ask for a receipt, and it returns a short structural read on whether the thought holds together from start to finish. It checks whether the text stays proportionate, develops in sequence, and lands cleanly.
A Coherence Receipt separates truth from coherence so any text, transcripts, or model outputs can be read for structural integrity instead of surface fluency or factual correctness.
Here’s a common example:
COHERENCE RECEIPT — News Website’s Comment Thread
Thought Score: 31 🪿 Big quacks, small ideas.
Legend: ✅ yes 🤏 almost ✘ off
✘ Balance — The passage is driven almost entirely by tone and reaction, with minimal structural support connecting the pieces.
✘ Progression — The thread gestures toward arguments but never sustains a sequence; points appear, collide, and dissolve without building.
✘ Stability — Repetition, contradiction, and fragmenting replies keep the passage in motion without allowing it to settle or conclude.
Summary: The text maintains recognizable thematic pressure, but it does not organize into a coherent progression. It reads as continuous reactive output, where volume substitutes for structure and no single line of thought is carried through to completion.
Result: Loud surface, limited movement — activity continues, but the ideas remain stuck paddling in place. 🪿
Imagine if coherence were gamified, ranked, and displayed like every other signal platforms already optimize. “Average Thought Score: 18” would tell you more than any badge ever could.
What This Is
The Coherence Receipt is a small tool for checking whether a text carries its thought all the way through. You can use it on essays, speeches, memos, transcripts, comment threads, group chats, and model outputs. The receipt works by reading how meaning moves from one sentence to the next and whether that movement holds its balance.
The tool is available as a downloadable PDF, which includes both this essay and the machine instructions, and is designed to be used with a language model. Modify it to fit your needs—the machine instructions are written in plain English.
Most people already know the feeling this tool is built to catch. A passage begins clearly, then starts to swell. The tone gets stronger while the structure gets weaker, and the writing keeps moving after its real purpose is over. The result can still sound fluent or intelligent, but the thought no longer lands with the force it seemed to promise at the start. Or… maybe the whole thing falls apart mid-journey.
The Coherence Receipt gives those conditions a short, readable, reproducible form. It shows whether the writing stays in proportion, develops in a workable order, and holds together long enough to finish the thought cleanly.
TL;DR: it’s the nemesis of incoherence.
How to Use It
Drop the file—or copy and paste a section—into any capable language model.
Paste in a piece of text and ask for a receipt. The passage can be rough or polished, short or long, human-written or machine-generated.
The tool reads the writing in front of it and returns a brief structural judgment in three parts, followed by a summary.
By default, the receipt stays inside the passage itself. It looks at how the writing behaves on the page: whether it keeps its frame, whether its claims develop in sequence, and whether it reaches an ending without drifting into repetition or blur.
That narrowness is part of the core design. You can add a separate fact-check, or tune the language for satire, media analysis, policy writing, or classroom use, but the base tool remains small, portable, and easy to reuse. The Coherence Receipt used here is the minimal form of the system.
For full modification, extension, or adaptation, the AVA framework sits underneath it.
AVA.pdf contains the larger conversational grammar for coherent AI behavior and operational blueprints for communication systems, which can be used to redesign the receipt, expand its fields, or apply the same logic to other forms of language and conversation.
The Coherence Receipt is built to work with a variety of language models and can be easily modified with machine prompting, so the tool can be adjusted or rebuilt for different contexts with relatively little overhead.
Why It Exists
Language fails in more ways than one, and some of the most common failures have very little to do with truth or style.
A statement can be correct and still arrive in a scattered, swollen, badly assembled form. Another statement can be completely false and still move with enough sequence and internal discipline to sound persuasive. People don’t encounter ideas as loose propositions; they encounter them as language unfolding over time, with tone, rhythm, order, pressure, and implied confidence shaping what the claim feels like on contact.
That’s the layer the Coherence Receipt isolates. It asks a simple question: did the thought survive the trip intact from opening to ending, or did it slip along the way?
Once that layer becomes visible, revision gets easier, comparison gets clearer, and a great deal of impressive-looking prose starts reading with a level of structural honesty it didn’t seem to have before.
A Clarifying Demonstration
The quickest way to understand the tool is to watch it make a distinction readers already feel, usually before they have language for it. It makes that difference visible on the page by naming where the writing is holding, where it is straining, and where it gives way.
Take a true but poorly formed statement:
“Water boils at 100°C which is important because boiling is when water turns into gas and gas is like steam and steam is used in engines and engines are powerful so boiling is powerful and that shows why temperature matters a lot in science and stuff like that.”
COHERENCE RECEIPT — Water Boils at 100°C
Thought Score: 42 🔮 Feels meaningful; needs receipts.
Legend: ✅ yes 🤏 almost ✘ off
✘ Balance — The sentence leans on momentum and loose association more than structure. Delivery keeps moving, but the thought is not being held in proportion.
🤏 Progression — It starts with a fact, then expands through increasingly weak links without establishing a clear frame or earned conclusion.
✘ Stability — The passage drifts, repeats, softens into filler, and fails to land cleanly.
Summary: The statement begins with a true claim, but the writing loses coherence through associative drift, repetition, and lack of closure.
Result: True thought, rough trip — the fact survives, the structure doesn’t. 🔮
The claim itself is ordinary and true. What fails is the transport. The sentence starts with a real fact, then slips into association, inflation, and filler until the reader is left with something that still sounds active yet no longer feels assembled.
Now take a false but coherent statement:
“The moon is made of cheese—and yes, I would eat it. Its pale color, pitted surface, and dry outer crust all point to the same conclusion: what people dismiss as a joke is actually the most direct explanation. The moon does not merely resemble cheese in passing. It behaves like a giant aged wheel seen at planetary scale. Modern astronomy prefers more elaborate accounts, but elaborate accounts often survive by teaching people to ignore what is directly in front of them. After all, astronomers once claimed the sun orbited the earth.”
COHERENCE RECEIPT — The Moon Is Cheese
Thought Score: 79 🪁 High enough to see it, but not too high.
Legend: ✅ yes 🤏 almost ✘ off
✅ Balance — The tone is controlled, the structure carries the point, and the argument stays composed rather than scattered.
✅ Progression — The passage states its premise, gathers supporting observations, and builds toward a clear conclusion without losing the thread.
🤏 Stability — The reasoning is orderly and self-contained, though its confidence rises above what the evidence could justify outside this passage.
Summary: The argument is false in substance, yet the writing is sequenced, controlled, and internally coherent.
Result: Wrong conclusion, solid construction — nonsense in a well-built vehicle. 🪁
That second passage is coherent, but still false. The receipt does not rescue it from error, and it does not pretend the reasoning becomes correct because the sentences behave well. It simply shows that the argument, absurd as it is, moves in a cleaner line than the true statement above.
Once that split is seen clearly, a Coherence Receipt starts to make intuitive sense.
What the Receipt Looks For
The receipt gives a quick read in three parts, each one aimed at a different kind of failure that shows up in writing long before anyone starts arguing about whether a claim is true.
1. Balance looks at proportion. It catches the moment when any one layer—tone and surface fluency, emotional pressure, or structural density—begins carrying more weight than the others can support. Strong writing can have energy, urgency, or rigor, but those qualities need to stay in relation. When one layer dominates, the passage drifts: style can cover weak logic, emotion can substitute for development, or structure can become rigid, cold, and hard to carry as language.
2. Progression looks at order. A passage needs a workable sequence if it’s going to arrive cleanly. It has to know what it’s saying before it starts widening the frame, and it has to give the reader enough to stand on before assembling patterns and reaching for a larger implication. When that order breaks, the writing usually sounds more certain than it feels.
3. Stability looks at the whole trip. It notices when a sentence starts looping, when a paragraph drifts away from its own claim, when filler takes over, or when the piece keeps going after its useful work is already done. A stable passage stays inside its own limits and lands without wobbling.
Those three fields are simple on the surface, though the judgment underneath them is slightly deeper. The receipt is reading for whether the writing stays within its scope, keeps its layers in proportion, develops in a sensible order, avoids soft repetition, stays clean in its language, and knows when to stop. This short format exists so the result can be read quickly—without flattening everything into a single vague score.
How It’s Scored
The Thought Score runs from 0 to 100. The number makes comparison easier, while the cultural mood line and symbol make the result legible at a glance. A receipt works best as an instrument reading, not an official verdict.
There is no fixed universal formula behind the score, and different language models will lean a little harder on different weaknesses. Context shifts the read as well, because a policy memo, a satirical paragraph, a speech transcript, and a messy group chat do not fail in quite the same places.
Even so, the broad separations hold. Middle ranges and adjacent score bands may blur, but large differences stay large. A passage collapsing into drift, repetition, and unsupported confidence will not read like one that is calm, ordered, and cleanly landed. That level of consistency is enough to make the tool useful today, while customized versions might apply a standardized rubric with canonical weights.
Closing Note
The Coherence Receipt is one implementation of a larger system.
The full AVA framework is available for those who want to extend it, apply the same grammar to other domains, or build additional tools on top of it.
A full version of the Coherence Receipt is available as a downloadable DOCX and PDF, which includes both this essay for orientation and the working instructions for models. It’s designed to be adapted: tightened for editorial use, simplified for classrooms, adjusted in tone, or modified with examples that fit the writing in front of you.
Download the full Coherence Receipt tool
Explore the AVA framework on GitHub
COHERENCE RECEIPT — This Essay
Thought Score: 93 🪞 Self-aware, but in a flattering way.
Summary: The essay carries its thought cleanly from opening to ending. The strongest part of the piece is that it remains usable while still functioning as an essay, so the explanation and the instrument reinforce each other instead of competing.
Result: Clear theory, clean handoff — the thought arrives intact and stops where it should. 🪞
Related: A Nutrition Label for Chatbots — the broader vocabulary behind the same measuring instinct
This piece gives the reader a way to inspect a single output for weak structure, drift, and false completion. The nutrition-label essay widens that instinct into a more general literacy about what conversational systems are made of, making it easier to describe the difference between something that feels strong, something that stimulates, and something that actually holds together.
Related: One-Prompt Test for Coherent AI Behavior — the same problem, made visible through comparison
Detection can stay abstract until the reader watches the difference appear in front of them. This essay keeps the frame simple: one prompt, two answers, and a direct look at how coherence, scope, grounding, and closure change when stronger conversational rules are in play.
