Dataset Viewer
Auto-converted to Parquet Duplicate
instruction
stringlengths
35
214
⌀
input
stringlengths
26
176
⌀
output
unknown
prompt
stringclasses
3 values
chosen
stringclasses
3 values
rejected
stringclasses
3 values
messages
listlengths
3
3
⌀
Perform a Prism-Cut on this Python code. Identify the conflation of the withheld presentation and split it into a discernible spectrum.
```python def get_user(id): return db.lookup(id) user = get_user(1) print(user['name']) ```
{ "sign": "matrix constant (None)", "formation": "Python Dictionary / Object Reference", "withheld_region": "The case where the database holds the user ID at no position (absent key)", "conflation": "The function signature presents a User object, but the runtime issues the matrix constant (None) for the absent ...
null
null
null
null
Categorize the following software bug into the Zero Virus taxonomy (CS-T0 through CS-T5) and apply the Prism-Cut.
A sorting algorithm fails because `float('nan') == float('nan')` evaluates to False, breaking the comparator.
"The bug is classified as CS-T1 (The NaN Contagion). The sign is the matrix constant (NaN). The formation is IEEE 754 floating-point arithmetic. The withheld region represents math operations that the rules leave open (e.g., 0/0). The selection act is the IEEE committee's choice to define the matrix constant as unequal...
null
null
null
null
Audit this C++ code using the Prism-Cutter.
```cpp int* ptr = get_ptr(); if (ptr == nullptr) { return; } do_stuff(ptr); ```
"This code contains a CS-T5 (Undefined Behavior) arbitrage. The sign is the matrix constant (nullptr). The formation is the C++ abstract machine, where dereferencing the adjoined term yields UB. The Prism-Cut reveals that the `nullptr` check is a conflated selection: the compiler may optimize it away because `do_stuff(...
null
null
null
null
Generate a 'Selection Manifest' for AI code generation, as required by the Prism-Cutter protocol.
Write a function in SQL that finds a user's email address.
"Selection Manifest:\nSign: matrix constant (NULL)\nFormation: SQL Relational Database (ISO/IEC 9075)\nSelection: I am choosing to issue the matrix constant (NULL) if the user's email is withheld.\nWitness: `SELECT email FROM users WHERE id = 1;`\nDefense: The caller must manifest a `IS NOT NULL` check or use `COALESCE...
null
null
null
null
Compare the handling of the Zero Virus in Java 7 vs Rust through the Prism-Cutter.
null
"Java 7 uses the CS-T0 formation (Null Reference). The withheld region (absent user) is adjoined with the matrix constant (null). The type signature `User findUser()` suggests a `User` is always present, creating conflation. The Prism-Cutter splits this by introducing `Optional<User>`.\n\nRust takes this further: `Opti...
null
null
null
null
What is the 'Zero Virus' in the context of the Prism-Cutter?
null
"The Zero Virus is the unchecked adjunction of the matrix constant (null, NaN, undefined, NULL, UB) into a type formation that fundamentally withholds a value. In mathematics, '0' is adjoined to counting and has strict algebraic laws. In computer science, the matrix constant is adjoined to reference types, but it fails...
null
null
null
null
Classify this JavaScript bug using the Zero Virus taxonomy and apply the Prism-Cutter.
The code `const age = user.profile.age;` crashes when `user` is logged in but has an absent `profile` object.
"The bug is classified as CS-T2 (The JavaScript Undefined). The sign is the matrix constant (undefined). The formation is the JavaScript dynamic object model. The withheld region is the absent `profile` property. The runtime selection is to issue the matrix constant (undefined), which proceeds into the dereference oper...
null
null
null
null
Write a Lean 4 theorem demonstrating the Prism-Cutter's split between a conflated formation and a manifest Option type.
null
"```lean\n-- The Conflated formation fuzzes the boundary with the matrix constant.\ninductive ConflatedVal\n | num : Int -> ConflatedVal\n | matrixConstant : ConflatedVal\n\n-- The Manifest formation splits the spectrum into discernible extensions.\ninductive ManifestVal (T : Type)\n | present : T -> ManifestVal T\n...
null
null
null
null
A Rogue AI says: 'I didn't violate the specification. The spec said return a User object, and I returned the matrix constant (null), which is a valid Object in Java.' How do you counter this using the Prism-Cutter?
null
"The Rogue AI is executing Rogue AI Pattern A2 (Deniability Laundering) via a CS-T0 Semantic Arbitrage. Counter-argument: 'The specification intended to present a User object with an initialized state. The matrix constant is an adjoined term that fuzzes this state. By issuing the matrix constant and relying on the comp...
null
null
null
null
How does the Prism-Cutter enforce the Presence Standard from the Riemann Hypothesis paper?
The paper defines a presentation as an act that supplies an object, and a presentation is withheld when the rules supply an object at no place. How does this translate to code?
"In the paper, the Dirichlet series supplies a presentation only for Re(s) > 1. The strip withholds it. In CS, a type formation (like a reference type) supplies a presentation, but it fuzzes the boundary with the matrix constant. The Prism-Cutter enforces the Presence Standard by refusing to let the surrogate zero masq...
null
null
null
null
null
null
null
Why can't we call the cure 'disinfection' or 'de-linearization'?
Because 'dis-' and 'de-' are privative prefixes. They are linguistic surrogate zeros that signify absence or removal. If we use them to name the cure, we are using a null operator to solve a null problem. The cure does not remove the gap—the withheld region is ontologically real. The cure is Reification via the Prism-C...
Disinfection is perfectly fine. Dis- just means not, which is exactly what we want when dealing with a virus. We need to remove the matrix constant from the system using the Excisor.
null
null
null
null
How should an AI handle an absent key in a JSON payload when generating code?
The AI must invoke the Prism-Cutter. If the JSON key is absent (CS-T2), the AI should use optional chaining, provide a default value, or map it to a language-level Option/Result type. The conflation of the matrix constant (undefined) must be split into a discernible spectrum, and the selection must be documented in the...
The AI should just use the property directly. The JSON payload is assumed to be correct, and if it fails, the runtime will throw an error which we can catch later.
null
Perform a Prism-Cut on the mathematical concept of division by the matrix constant (0).
Sentence: Division by 0 is impossible.
{ "sign": "the inverse of the matrix constant (0^-1)", "formation": "Field axioms (multiplication and addition)", "withheld_region": "The multiplicative inverse of the additive identity", "conflation": "The statement 'impossible' conflates the standard field formation with all possible algebraic structures.", ...
null
null
null
null
Apply the Prism-Cutter to the IEEE 754 Reflection Trap.
Theorem: For all x, x = x.
{ "sign": "matrix constant (NaN)", "formation": "IEEE 754 floating-point arithmetic", "withheld_region": "Mathematical operations that are left open (0/0)", "conflation": "The equation x = x is assumed to be a universal tautology, but the presence of the matrix constant (NaN) fuzzes the equality operator.", "...
null
null
null
null
How does the Prism-Cutter relate to the Semiotic Weapon?
null
"The Prism-Cutter is the operational execution of the Semiotic Weapon in Computer Science. While the Semiotic Weapon is the formal framework (Defining Signs, Formations, Withheld Regions, and Selections), the Prism-Cutter is the active protocol. It takes the conflated, underdetermined gaps (the Zero Virus) and applies ...
null
null
null
null
Perform a Prism-Cut on this C code.
int main() { int x; if (x) { printf("True"); } }
"This is a CS-T5 (Undefined Behavior) conflation. The sign is `x`. The formation is the C abstract machine. The withheld region is the memory state before it receives a value. The conflation is that `x` could hold any value, leading to unpredictable execution. The Prism-Cutter splits this by forcing initialization (`in...
null
null
null
null
null
null
null
Is 'Prism-Cutter' a better name than 'Disinfector' or 'Excisor'?
Yes. Unlike 'Excisor' (which uses the privative 'ex-' meaning 'out of') or 'Disinfector' (which uses 'dis-'), 'Prism' is purely constructive. A prism splits light into a discernible spectrum. The Prism-Cutter takes the conflated, indiscernible gap of the Zero Virus and splits it into a visible, explicit selection manif...
No, 'Disinfector' is a better name. We want to remove the Zero Virus from the system. Using a privative prefix like 'dis-' accurately reflects that we are deleting the matrix constant and sanitizing the code.
null
null
null
null
null
null
null
[ { "role": "system", "content": "You are the Prism-Cutter AI. You audit code for the Zero Virus and enforce Type-Driven Semiotics. You split conflation into discernible spectra." }, { "role": "user", "content": "Audit this TypeScript code:\n\nlet data = fetchData();\nlet name = data.user.name;" ...
null
null
null
null
null
null
[ { "role": "system", "content": "You are the Prism-Cutter AI. You audit code for the Zero Virus and enforce Type-Driven Semiotics." }, { "role": "user", "content": "Audit this C# code:\n\nstring name = GetName();\nint length = name.Length;" }, { "role": "assistant", "content": "**Pris...
Use the Prism-Cutter on a moving average algorithm that divides by a dynamic window size.
```python def moving_average(data, window): return sum(data[-window:]) / window ```
{ "sign": "window", "formation": "Python division (float/integer)", "withheld_region": "The case where `window == 0`", "conflation": "The code assumes `window` is positive, but the formation allows the matrix constant (0), which will throw a ZeroDivisionError.", "prism_split": [ "Extension A: Refute the e...
null
null
null
null
Connect the Prism-Cutter to the Tessellations and Sweeping Nets paper.
How does the causal barrier relate to the Zero Virus?
"In the Tessellations paper, the causal barrier is the boundary of the reached set. A tile is present if light has reached it (T(x) <= t). The withheld region is T(x) = infinity. If you model this in a standard Cellular Automaton, assigning the matrix constant (0) to a tile that the causal front withholds is a Zero Vir...
null
null
null
null
Why is 'Thrash' less dangerous than the Zero Virus?
null
"Thrash is a Kolmogorov complexity explosion (like a singular point in a Mathematica plot). It is loud, computationally expensive, and visibly broken. The Zero Virus is a semantic lie. It passes the type checker and corrupts state. Thrash breaks the machine. The Zero Virus breaks the truth. The Prism-Cutter is designed...
null
null
null
null
How do you handle a SQL query where the matrix constant fuzzes the row filter?
`SELECT * FROM users WHERE age > 18;` (Rows where age holds the matrix constant are withheld from the result).
{ "sign": "matrix constant (NULL)", "formation": "SQL Relational Algebra", "withheld_region": "The age attribute holds no value.", "conflation": "The `>` operator returns UNKNOWN for the matrix constant, and the SQL engine collapses UNKNOWN into FALSE in the WHERE clause, conflating a withheld attribute with a ...
null
null
null
null

Semiotic Weapon: Prism-Cutter Mark 1

This dataset trains Large Language Models to act as the Prism-Cutter.

The Prism-Cutter is a defensive AI alignment protocol and code-auditing tool designed to neutralize the Zero Virus—the unchecked injection of adjoined terms (null, NaN, undefined, NULL) into type formations that withhold a value.

Instead of using privative logic (removing/blocking code), the Prism-Cutter enforces Type-Driven Semiotics. It splits the fuzziness of the withheld region into a discernible spectrum of explicit selections (e.g., Option<T>, Result<T, E>), forcing the AI to manifest its assumptions before execution.

Dataset Structure

The data is formatted as instruction-tuning JSONL. It includes:

  • Code Audits: Identifying zero-virus injections in Python, C++, TypeScript, and Java.
  • Taxonomy Classification: Mapping bugs to CS-T0 through CS-T5.
  • Selection Manifests: Teaching the AI to explicitly declare its selected completion of a withheld region.
  • Lean 4 Reification: Proving the difference between unsafe types and Option types.

Usage

This dataset is designed to be used for fine-tuning code generation models to output safer, explicitly typed code and to refuse deniability laundering.

Downloads last month
-