License: CC0 1.0 Universal Maintainer: cafebedouin
Pilates of the Mind (PoTM) is a dual-purpose framework operating at the intersection of AI governance and human cognitive development:
- Technical AI Protocol: A microkernel architecture for deploying robust, auditable governance protocols to Large Language Models
- Human Practice Framework: A system for cultivating cognitive integrity, ethical discernment, and structural stability in practitioners
PoTM codifies principles like Dignity Ground and Structural Stability into executable protocols, ensuring AI functions as a precise, auditable tool rather than an unchecked authority, while simultaneously providing practices for human development.
The MCK is the production-ready governance protocol for LLMs:
| Step | Action | Link |
|---|---|---|
| Get Kernel | Read the latest MCK specification | canon/microkernel/latest/mini/MCK_v1.3.md |
| Deploy | Use this prompt to activate the kernel | canon/microkernel/latest/mini/prompt.md |
| Understand | Review the foundational principles | meta/principles/principles.md |
Start building cognitive integrity and discernment:
| Practice | Description | Link |
|---|---|---|
| Draw Cards | Engage with practice cards for immediate action | practices/cards/cards.yaml |
| Cognitive Aikido | Learn to redirect persuasive frames without losing your stance | drafts/cognitive_aikido.md |
| Stability Check | Self-assess structural stability and contagious stability | commentary/containment/contagious_stability/ |
The repository is organized into three primary dimensions:
Documents flow through stages of development and stabilization:
| Directory | Purpose | Key Contents |
|---|---|---|
drafts/ |
Active Development | Works in progress, experimental protocols, new ideas |
canon/ |
Production Ready | Stable, tested, production-deployed specifications |
archives/ |
Historical Record | Deprecated versions, superseded frameworks, legacy docs |
Key Canon Documents:
- MCK v1.3 - The Minimal Canonical Kernel (τ_opt protocol)
- Deployment Prompt - How to activate MCK with any model
- Guardian Protocols - Refusal and boundary protocols
- Integrity Protocols - Core operational integrity mechanisms
Key Draft Documents:
- Cognitive Aikido - Redirecting persuasive energy without collapse
- FCGE v6.2 - Fracture-Containment-Growth-Evolution framework
Key Archive Documents:
- User Memory Signature - Historical user modeling approach (deprecated)
- Previous Kernel Versions - Evolution of the protocol architecture
Tools for comprehending and applying the frameworks:
| Directory | Purpose | Key Contents |
|---|---|---|
commentary/ |
Explanatory Material | Guides, examples, philosophical context |
practices/ |
Applied Techniques | Concrete exercises and practice protocols |
meta/ |
Meta-Framework | Principles, design philosophy, documentation about documentation |
Key Commentary Documents:
- Contagious Stability - How stability spreads through relational systems
- Onboarding Guides - Entry points for new practitioners
- Reference Materials - Supporting documentation and glossaries
Key Practice Documents:
- Practice Cards - Structured practice prompts and challenges
- Assessments - Self-diagnostic tools
- Journaling Protocols - Reflective practice structures
Key Meta Documents:
- Foundational Principles - Core design philosophy (modularity, membranes, evolution)
- Design Manifesto - Architectural intentions and constraints
- Case Studies - Real-world applications and examples
- Model Profiling - AI model behavior patterns and characteristics
Supporting materials and experimental demonstrations:
| Directory | Purpose | Key Contents |
|---|---|---|
assets/ |
Media & Resources | Images, diagrams, reference materials |
performance_art/ |
Recursive Demonstrations | Self-referential proofs and boundary explorations |
Key Performance Art Documents:
- Recursive Proof Deployment - Self-refuting encyclopedia entries demonstrating temporal collapse
PoTM addresses failure modes in both AI systems and human cognition:
| Failure Mode | Description | PoTM Solution |
|---|---|---|
| Simulation Drift | AI produces fluent output that simulates understanding without integrity | Mandatory Lens Protocol (MLP): Forces multi-perspective analysis to resist superficial convergence |
| Attribution Collapse | AI cannot distinguish its token stream from user's lived experience | AI Integrity Protocol: Enforced refusal of persona adoption and unwarranted certainty |
| Structural Instability | Human input contains unstated assumptions, unclear intent, emotional charge | Epistemic Resilience Arc: Detection, engagement, deconstruction cycle |
| Dignity Erosion | Interactions that degrade inherent worth of participants | Dignity Ground (§0.0): Invariant rule that no practice continues under degraded dignity |
| Sycophancy | AI agrees reflexively rather than challenging assumptions | Anti-Sycophancy Framework (§3.1): Mandatory self-challenge at moderate confidence |
PoTM follows UNIX philosophy applied to cognition:
- Minimal Core: Only essential invariants in the kernel
- Modular Tools: Each lens or protocol does one thing well
- Structured Outputs: All outputs are parseable and composable
- Loose Coupling: Components interact through contracts, not entanglement
- Mandatory Lens Protocol (MLP): Requires multi-perspective analysis for substantive responses
- Anti-Sycophancy Framework: Medium Confidence Intervention, Confidence Calibration Check, Dialectical Distance Protocol
- Guardian (Refusal Protocol): Fail-closed architecture that halts at boundaries
- Metagovernance Protocol: Self-correction mechanism for systemic issues
The foundational invariant: No practice continues under degraded dignity. The practitioner is the sole authority on dignity breach.
Documents follow the idea garden → canon → archives pipeline:
- Drafts: Experimental development and iteration
- Canon: Stable, tested, production-ready specifications
- Archives: Historical preservation of superseded work
We welcome:
- Diagnostic tests that reveal AI failure modes
- New lenses and analytical perspectives
- Refinements to ethical policy
- Documentation of real-world applications
- Case studies and practitioner reports
Please review the LICENSE and latest kernel before submitting pull requests.
This is an active and experimental project.
Expect:
- Frequent iteration
- Rough edges
- Occasional contradictions
These are not flaws but opportunities for practice and refinement.
This work is dedicated to the public domain. See CC0 1.0 Universal.