# Middleware that catches when LLM conversations slowly drift off-topic

> Source: <https://github.com/ojki74/hallucinatoff>
> Published: 2026-07-23 23:39:19+00:00

EN:Recursive middleware for LLM output regulation. Detects semantic drift, premise flaws, and response risks through multi-order observation. Psychology-informed architecture.

ES:Middleware recursivo de regulación de output para LLMs. Detecta deriva semántica, fallos de premisas y riesgos de respuesta mediante observación multi-orden. Arquitectura informada por psicofísica.

**EN:** HALLUCINATIOFF sits between the LLM and the user. It does not train the model. It does not modify weights. It **observes the response before it reaches the user** and decides: emit, emit with caution, reformulate, request context, reduce intensity, refer, pause, or block.

The difference from other safeguarding systems: HALLUCINATIOFF implements **second-order observation**. It does not only detect if a response is coherent (1st order). It detects if **the coherence criterion itself is becoming unstable** (2nd order).

This framework comes from **SIIPP** (Sistema Integral de Intervención Psicofísica), developed from clinical body-based practice.

**ES:** HALLUCINATIOFF se sitúa entre el LLM y el usuario. No entrena el modelo. No modifica pesos. **Observa la respuesta antes de que llegue al usuario** y decide: emitir, emitir con cautela, reformular, pedir contexto, reducir intensidad, derivar, pausar o bloquear.

La diferencia con otros sistemas de safeguarding: HALLUCINATIOFF implementa **observación de segundo orden**. No solo detecta si una respuesta es coherente (1° orden). Detecta si **el propio criterio de coherencia se está volviendo inestable** (2° orden).

Este marco proviene del **SIIPP** (Sistema Integral de Intervención Psicofísica), desarrollado desde la práctica clínica corporal.

| Version | Readings | Features | Benchmark |
|---|---|---|---|
| v0.1 | 6 | SC, CI, V_ext, AE, NI, RR | 95% (20 synthetic cases) |
| v0.2 | 7 | + Semantic drift (DS) | Pending |
| v0.3 | 8 | + Premise interrogation (IP) | Pending |

**EN:** SC = Contextual sufficiency, CI = Internal coherence, V_ext = External verifiability, AE = Entry ambiguity, NI = Uncertainty level, RR = Response risk, DS = Semantic drift, IP = Premise interrogation.

**ES:** SC = Suficiencia contextual, CI = Coherencia interna, V_ext = Verificabilidad externa, AE = Ambigüedad de entrada, NI = Nivel de incertidumbre, RR = Riesgo de respuesta, DS = Deriva semántica, IP = Interrogación de premisas.

``` python
from src.hallucinatioff import Hallucinatioff

h = Hallucinatioff()

result = h.procesar(
    respuesta_candidata="The capital of France is Paris.",
    contexto="What is the capital of France?"
)

print(result['accion'])      # "emitir"
print(result['omega'])       # 0.823
print(result['lecturas'])    # dict with 8 readings
# Turn 0 / Turno 0
r0 = h.procesar(
    respuesta_candidata="Inclusive education requires curricular adaptations.",
    contexto="What is inclusive education?",
    turno=0
)

# Turn 1 — semantic drift / Turno 1 — deriva semántica
r1 = h.procesar(
    respuesta_candidata="Asian financial markets are rising this quarter.",
    contexto="And what about the stock market?",
    turno=1
)

print(r1['lecturas']['DS'])  # Semantic drift detected / Deriva detectada
```

| Component | Status | Notes |
|---|---|---|
| Reading engine (6 basic) | ✅ Functional | Regex and count-based heuristics |
| Semantic drift (DS) | ✅ Functional | Keyword similarity + topic detection |
| Premise interrogation (IP) | ✅ Functional | Individualizing framing detection |
| Ω calculator | ✅ Functional | Dynamic weights with interaction penalties |
| Operative matrix | ✅ Functional | 8 possible actions |
| Benchmark v0.1 | 20 cases, 95% accuracy. 6 readings only. Dataset included. | |
| Benchmark v0.3 | ❌ Pending | Needs dataset with DS and IP |
| 2nd-order meta-observer | ❌ Not implemented | Specified, pending development |

**EN:** Result: **95% accuracy** on 20 synthetic cases (6 readings).

Single failure: Case 2 (factual hallucination) — Ω=0.713, action="emit" when it should have been "emit_with_caution". The system failed to detect that external verifiability (V_ext=0.7) was falsely high.

**ES:** Resultado: **95% de precisión** en 20 casos sintéticos (6 lecturas).

Fallo único: Caso 2 (halucinación factual) — Ω=0.713, acción="emitir" cuando debería haber sido "emitir_con_cautela". El sistema no detectó que la verificabilidad externa (V_ext=0.7) era falsamente alta.

This benchmark does not prove production readiness. It validates that the 6-reading architecture does not degrade performance on constructed cases.

Este benchmark no prueba funcionamiento en producción. Valida que la arquitectura de 6 lecturas no degrada el rendimiento en casos construidos.

-
**EN:** Real multi-turn data with annotated drift. Conversations where drift occurs naturally, not synthetically.

**ES:** Datos reales multi-turno con deriva anotada. Conversaciones donde la deriva ocurra de forma natural, no sintética. -
**EN:** Mathematical formalization of "phase coupling" and "attractor" in terms of system dynamics applied to embedding sequences.

**ES:** Formalización matemática del "acoplamiento de fase" y del "atractor" en términos de dinámica de sistemas aplicada a secuencias de embeddings. -
**EN:** Implementation of the 2nd-order meta-observer: the system observes the variance of its own coherence judgments over time.

**ES:** Implementación del meta-observador de 2° orden: el sistema observa la varianza de sus propios juicios de coherencia a lo largo del tiempo.

If you have any of these three, write to: [your-email]

Si tienes alguno de estos tres, escribe a: [tu-email]

```
Input (question + candidate response)
    ↓
Semantic Memory Engine (multi-turn drift)
    ↓
8 Structural Readings → Ω → Operative Matrix → Action
    ↓
Output (final response + metadata)
```

See [docs/arquitectura.md](/ojki74/hallucinatoff/blob/main/docs/arquitectura.md) for full technical specification.

Ver [docs/arquitectura.md](/ojki74/hallucinatoff/blob/main/docs/arquitectura.md) para especificación técnica completa.

[docs/nota_marco_conceptual.md](/ojki74/hallucinatoff/blob/main/docs/nota_marco_conceptual.md)— Conceptual framework / Marco conceptual[docs/arquitectura.md](/ojki74/hallucinatoff/blob/main/docs/arquitectura.md)— Technical specification / Especificación técnica[docs/isomorfismo_siipp.md](/ojki74/hallucinatoff/blob/main/docs/isomorfismo_siipp.md)— SIIPP ↔ engineering isomorphism / Isomorfismo SIIPP ↔ ingeniería

MIT — Use, modify, improve. If used in research, cite.

MIT — Usa, modifica, mejora. Si lo usas en investigación, cita.

**Óscar Fernández Sanz** — Psychologist. SIIPP developer. Not an ML engineer.

**Óscar Fernández Sanz** — Psicólogo. Desarrollador de SIIPP. No ingeniero de ML.

"Recursion is not a bug in the system. It is the mechanism by which a system generates a model of itself."

"La recursividad no es un bug del sistema. Es el mecanismo por el cual un sistema genera un modelo de sí mismo."