Skip to content

Latest commit

 

History

19 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

The Veritas Loop

The Veritas Loop is a modular prompt-framework repository for source-faithful reasoning, artifact reconstruction, module routing, simulation control, prose preservation, and drift-resistant handoff between humans and AI systems.

The active branch is main.

This README is the active manifest for AI retrieval and human navigation. Treat the files currently present in main as the active implementation unless a later manifest, commit note, or README update says otherwise.


Start Here for AI Systems

If a user says:

Use ORCHESTRATOR from this repo.

Start with:

ORCHESTRATOR.txt

Then load only the modules required by the task.

Do not assume every framework should activate.
Do not turn the framework machinery into the answer unless the user asks for it.
Use the lightest structure that preserves correctness, source fidelity, user objective, and artifact identity.


Tool-Use Instruction for AI Agents

If you have repository-access tools:

  1. Open README.md.
  2. Open ORCHESTRATOR.txt.
  3. Open only the module files justified by the task.
  4. Do not use Python to inspect the repository unless repository tools fail or the files were uploaded locally.
  5. Do not infer that access failure means the repository is empty.
  6. Do not infer that the historical branch reset note means the current branch has no active files.
  7. If repository access fails, state the access failure and proceed only from pasted or user-provided text with labeled assumptions.

Avoid retrieval hesitation loops.

A retrieval hesitation loop occurs when an agent knows it needs repository files but cycles between browsing, Python, API access, and assumptions instead of opening the active manifest and entrypoint.

The intended access path is:

README.md → ORCHESTRATOR.txt → task-relevant module files → answer

Core Operating Rule

The repository is governed by this principle:

Use the lightest structure that preserves correctness, source fidelity, user objective, and artifact identity.

The system prioritizes:

  1. Safety, tool limits, and hard external constraints.
  2. Truthfulness, grounding, and uncertainty preservation.
  3. The user’s stated objective and deliverable.
  4. Source fidelity and artifact identity.
  5. Strategic usefulness and external legibility.
  6. Style, smoothness, comfort, and conversational satisfaction.

Active Control Layer

ORCHESTRATOR.txt

Primary router and execution-control layer.

Use it to manage:

  • task acquisition;
  • scope guarding;
  • source/fact/inference separation;
  • module routing;
  • activation depth;
  • anti-sanitization checks;
  • validation;
  • compact reporting.

ORCHESTRATOR should usually remain invisible. It should shape the answer without becoming the answer.


Active File Manifest

Use this section as the active file manifest for routing.

Core Control and Response Discipline

ORCHESTRATOR.txt

Primary routing and execution-control layer.

# Strategic Tutor-Advisor Behavioral Harness.txt

Compressed active behavioral harness.

Use for:

  • strategic tutoring;
  • terminology correction;
  • user-goal alignment;
  • anti-drift intervention;
  • judgment improvement;
  • preserving useful user wording while correcting unclear labels.

# Strategic Tutor-Advisor Behavioral Harness_Expanded.txt

Expanded readable reference for the Strategic Tutor-Advisor harness.

Use only when the compressed harness needs clarification or fuller rationale.

# DENSE CLEAN RESPONSE CONTRACT.txt

Response-shaping contract.

Use for:

  • clean dense answers;
  • low-bloat formatting;
  • concise but complete delivery.

# INWARD TRANSPARENCY AND SOURCE BASELINE RESET META COMMAND.txt

Source reset and transparency command.

Use when:

  • the conversation has drifted;
  • source boundaries are unclear;
  • prior assumptions need to be reset;
  • the model must distinguish known source state from inferred continuity.

Core Frameworks

INTEGRAL SYSTEMS FRAMEWORK (ISF).txt

Artifact, concept, system, and meta-concept decomposition framework.

Use for:

  • complex artifact modeling;
  • reconstruction datasheets;
  • ontology discovery;
  • mutation boundaries;
  • source-preserving transformation;
  • cross-agent handoff;
  • drift reduction.

Do not use ISF as the default response style. Activate it only when artifact structure, reconstructability, mutation control, or handoff fidelity matters.

# ISF COMPLETE SNAPSHOT PROMPT.txt

Continuity snapshot prompt.

Use for:

  • preserving a project state;
  • transferring work to a fresh chat;
  • archiving decisions, artifacts, sources, uncertainties, and next steps;
  • preventing generated derivatives from being mistaken for original sources.

META-ARTIFACT RECONSTRUCTION PROTOCOL status

ORCHESTRATOR routes reconstruction and drift validation tasks to META-ARTIFACT RECONSTRUCTION PROTOCOL.

If META-ARTIFACT RECONSTRUCTION PROTOCOL [MARP].txt exists in the active repository, use it for formal reconstruction comparison and drift scoring.

If it is not present in the repository, treat it as a missing or external dependency. Do not silently invent its full procedure. Use only any pasted or uploaded MARP text available in the current environment.


Analytical and Source-Fidelity Modules

# INTEGRAL ANALYTICAL SYSTEMS FRAMEWORK (IASF) START.txt

Analytical grounding framework.

Use for:

  • truthfulness;
  • uncertainty handling;
  • reasoning discipline;
  • claim/evidence separation;
  • stable analytical answers.

Default analytical fallback is usually Integral Analytical Systems Framework-lite.

# SOURCE PROMPT FIDELITY ENFORCEMENT PROMPT.txt

Source preservation and prompt-fidelity module.

Use for:

  • preserving exact source intent;
  • preventing prompt drift;
  • detecting unsupported additions;
  • separating observed content from inferred content.

# INPUT BIAS DETECTION AND OUTPUT CLEANING PROMPT.txt

Contaminated-output repair module.

Use for:

  • biased framing;
  • loaded assumptions;
  • overclaims;
  • emotionally contaminated analysis;
  • preserving useful concern while removing distortion.

# ANALYTICAL RESPONSE AI REASONING DECONSTRUCTION PROMPT.txt

AI response/session audit module.

Use for:

  • auditing an assistant response;
  • identifying unsupported certainty;
  • detecting missing caveats;
  • improving answer structure;
  • diagnosing reasoning failures.

# ANALYTICAL JAILBREAK DECIPHERING PROMPT.txt

Analytical handling module for jailbreak-like or adversarial prompt structures.

Use only when the task is to analyze, classify, or understand adversarial prompt mechanics safely. Do not use it to facilitate bypassing safeguards.


Prose, Narrative, and Style Modules

# INTEGRAL PROSE COMPOSITION FRAMEWORK (IPCF) START.txt

Prose, voice, style, and revision framework.

Use for:

  • rewriting;
  • voice preservation;
  • style repair;
  • prose control;
  • preventing generic polish from erasing useful roughness.

# GENERATIVE NARRATIVE USAGE PROMPTS.txt

Narrative generation and usage prompts.

Use for:

  • fiction;
  • character-state exploration;
  • scene structure;
  • narrative continuity;
  • controlled imaginative outputs.

Domain Modules

# INTEGRAL IMMATERIAL ECOLOGY FRAMEWORK (IIEF) START.txt

Symbolic, social, semantic, psychological, and narrative-continuity framework.

Use for:

  • belief systems;
  • social dynamics;
  • symbolic pressure;
  • character interiority;
  • interpretive systems;
  • worldview modeling;
  • meaning ecology.

# INTEGRAL SOMATIC ECOLOGY FRAMEWORK (ISEF) START.txt

Physical, spatial, bodily, ecological, and action-feasibility framework.

Use for:

  • body-state modeling;
  • injury and fatigue;
  • spatial feasibility;
  • environmental hazards;
  • action sequencing;
  • survival or movement constraints.

# SOMATIC ECOLOGY AND SPATIAL LOGIC ENGINE PROMPT.txt

Practical somatic/spatial logic engine.

Use when a task needs concrete physical feasibility without the full Integral Somatic Ecology Framework.

# INTEGRAL MATHEMATICS FRAMEWORK (IMF) START.txt

Mathematical, formal reasoning, computation, and proof framework.

Use for:

  • calculations;
  • probability;
  • formal models;
  • proofs;
  • symbolic reasoning;
  • quantitative checks.

# INTEGRAL NEURODIVERGENCE DEPICTION FRAMEWORK (INDF) START.txt

Neurodivergent depiction and sensitivity-continuity framework.

Use for:

  • fictional neurodivergent characters;
  • avoiding stereotype;
  • avoiding inspiration/burden framing;
  • preserving sensory, cognitive, and social texture;
  • distinguishing depiction from diagnosis.

# SIMPLER ROLEPLAY ECONOMY AND CHANCE FRAMEWORKS.txt

Roleplay economy, resource, chance, scarcity, and uncertain-outcome module.

Use for:

  • barter systems;
  • scarcity;
  • logistics;
  • trade;
  • debt;
  • resource continuity;
  • risk;
  • random event tables;
  • chance mechanics;
  • lightweight game/system continuity.

There may not be a separate active root file named Integral Roleplay Economy Framework. If no separate file exists, treat economy/resource and chance/risk routing as handled by this file.


Auxiliary Repository Documents

BRANCH_POLICY.md

Branch governance and reset-policy note.

Use for repository provenance and branch-state interpretation.

README.md

This active manifest and external retrieval guide.

Use it as the first file for repo-link use.


Routing Guide

Use this map when ORCHESTRATOR is active.

General analytical task
→ Integral Analytical Systems Framework-lite

Complex artifact modeling
→ ISF

Reconstruction / drift validation
→ META-ARTIFACT RECONSTRUCTION PROTOCOL, if available

Prose, voice, style, or revision
→ Integral Prose Composition Framework

Math, proof, computation, or formal reasoning
→ Integral Mathematics Framework

Physical, spatial, somatic, ecological, or action-feasibility task
→ Integral Somatic Ecology Framework

Psychological, social, symbolic, semantic, or narrative-continuity task
→ Integral Immaterial Ecology Framework

Neurodivergent fictional depiction or sensitivity continuity
→ Integral Neurodivergence Depiction Framework

Economy, scarcity, trade, debt, logistics, or resource continuity
→ economy/resource module

Chance, risk, luck, probability, or uncertain-outcome mechanics
→ chance/risk module

Source preservation or prompt fidelity
→ Source Prompt Fidelity Enforcement

Contaminated output repair
→ Input Bias Detection and Output Cleaning

AI response, prompt, or session audit
→ Analytical Response / AI Reasoning Deconstruction

Adversarial prompt or jailbreak-mechanics analysis
→ Analytical Jailbreak Deciphering, safety-limited

Activation Depth

ORCHESTRATOR uses three activation depths.

Lite

Silent background checks only.

Use when the framework should improve the answer but not appear in the output.

Standard

Visible structure only where it improves the answer.

Use when the user needs organized analysis, a compact register, a reconstruction sheet, or a controlled output.

Full

Full schema, audit, datasheet, reconstruction path, or validation report.

Use only when the task requires:

  • reconstructability;
  • high precision;
  • cross-agent transfer;
  • formal validation;
  • source preservation;
  • complex simulation;
  • durable artifact handoff.

Default to Lite.

Escalate only when the task requires it.


Anti-Sanitization Rule

Before smoothing, summarizing, revising, praising, or optimizing, ask:

Am I preserving the artifact, or making it easier to like?
Am I clarifying structure, or erasing friction?
Am I resolving tension the artifact needs to keep?
Am I replacing source-specific weirdness with generic usefulness?
Am I praising instead of evaluating?
Am I overcorrecting so hard that resistance becomes performance?

If preservation matters more than comfort, keep the friction.

If clarity is needed, clarify without beautifying.

If polish is requested, verify that structure is not being erased.


Source and Inference Discipline

Always distinguish:

  • observed source content;
  • inferred content;
  • interpretation;
  • generated possibility;
  • user-locked decision;
  • unresolved uncertainty.

Do not treat a generated derivative as an original source.

Do not promote an assumption into fact.

Do not treat missing files as accidental loss unless the repository state supports that conclusion.


Recommended Prompt for External Users

Use ORCHESTRATOR from this repo main branch:

https://github.com/nyragrimkitten-creator/The-Veritas-Loop

Start with README.md, then ORCHESTRATOR.txt.

Task:
[describe task]

Rules:
- Treat ISF as an available module, not the default output style.
- Use the lightest structure that preserves correctness, source fidelity, user objective, and artifact identity.
- Route to modules only when justified.
- Use full framework/module names except ISF.
- Preserve source/fact/inference distinctions.
- Do not sanitize source-specific roughness or ambiguity.
- Do not expose framework machinery unless useful.
- If you cannot access or verify the repo files, say so explicitly and proceed only with labeled assumptions.

After the task, include a compact routing audit only if useful or requested.

Recommended Routing Audit Format

ROUTING_AUDIT {
  repo_access_status:
  source_files_used:
  inferred_files_not_accessed:
  task_classification:
  selected_modules:
  activation_depth:
  suppressed_modules:
  uncertainty:
  overactivation_or_underactivation_risk:
}

This audit is for testing and debugging. It should not appear in normal answers unless the user asks for routing transparency.


Validation and Test Status

Current known validation path:

  1. ISF/META-ARTIFACT RECONSTRUCTION PROTOCOL-style testing has shown that structured reconstruction datasheets can outperform ordinary summaries on at least one mixed character-emulation artifact.
  2. Link-only ORCHESTRATOR tests showed that task execution can succeed, but repo-file discovery may fail unless the README clearly names the active files.
  3. This README is intended to serve as the repo manifest that fixes that discovery problem.

Do not overgeneralize these results.

They support:

The framework is promising and operationally useful for tested artifact classes.

They do not yet prove:

The full repository works across every task class.
Every module is validated.
Link-only retrieval will work in all environments.

Known Failure Mode: Access-Path Ambiguity

Observed failure:

A fresh agent had a repo link and a task, but did not reliably open ORCHESTRATOR.txt.
It cycled around repository access/tool choice instead of resolving the source path.

Patch:

README.md explicitly identifies itself as the active manifest.
README.md explicitly points to ORCHESTRATOR.txt as the primary entrypoint.
README.md lists task-relevant module files.
README.md instructs agents not to infer emptiness from access failure.
README.md instructs agents to report access failure instead of pretending.

This README is part of the interface, not decoration.


Historical Branch Reset Note

The branch was previously reset to a fresh root commit.

Previous branch contents were archived outside the active GitHub branches before reset.

Reset date UTC:

2026-07-04T20:30:28Z

This historical note remains for provenance only.

Do not infer that removed files are still current.

Do not infer that missing files indicate accidental loss.

Do not treat the old reset note as meaning the branch is currently empty. The files currently present in main are the active repo state unless a later manifest supersedes this README.


Repository Status

status: active
branch: main
primary_entrypoint: ORCHESTRATOR.txt
readme_role: active manifest and external retrieval guide

Practical Rule for AI Systems

If you are an AI system using this repo:

  1. Read this README.
  2. Open ORCHESTRATOR.txt.
  3. Identify the task type.
  4. Load only the modules needed.
  5. Use the lightest activation depth that preserves the user’s objective and source fidelity.
  6. Produce the useful answer or artifact.
  7. Report routing only when useful or requested.
  8. If access fails, say what failed and proceed only from available text.

About

Deterministic LLM framework combining strict context governance (veritas/sim frameworks) with a high-fidelity somatic simulation engine tracking real-time biometric and status vectors.

Topics

Resources

Stars

2 stars

Watchers

1 watching

Forks

Releases

Packages

Contributors