tscg

Filter skill for evaluating M2 and M1_CoreConcepts candidates — guards against Ontological Overfitting (adding one M2 concept per phenomenon instead of composing existing ones). Use this skill whenever a session proposes a new M2 GenericConcept or M1_CoreConcepts combo, or whenever a concept is described as "missing" from M2. Applies three sequential filter tests: Decomposability, Transdisciplinarity, Atomicity. Produces a placement decision: M2 / M1_CoreConcepts / M1_Domain / Rejected. See also: OntologicalOverfitting.md in CoreHypotheses/

Debugging and diagnostic skill for TriskeleToolchain (tsk-cc, tsk-link, triskele-vm) when a pipeline run fails or produces incorrect output. Use this skill whenever Michel uploads a pipeline_report.txt, a .tvmx binary, a .ll file, or describes a VM crash, wrong test result, or linker error. Also use when discussing multi-platform targets (Windows, Linux, WASM) or codegen correctness issues. NOT for new feature implementation (use tscg-rust-vm skill) and NOT for TSCG ontology work.

Skill for implementing TriskeleVM and the TriskeleToolchain in Rust. Use this skill whenever Michel asks to implement, extend, or debug any component of the TriskeleToolchain in Rust: the VM interpreter loop, ISA opcode execution, memory management (stack/heap/arenas), FFI bindings to SDL2/libc, Im_REGISTER_CB callbacks (C→Rust), the tsk-asm assembler, tsk-dis disassembler, tsk-link linker, or tsk-build orchestrator. Also use when creating Cargo.toml files, workspace structures, or when discussing Rust patterns specific to TriskeleVM (unsafe FFI, libloading, arena allocators, register-based interpreter dispatch). Do NOT use for BabylonJS simulations, Python scripts, or JSON-LD ontology work — those have their own skills.

Migration pipeline for the TSCG foundational formalism reform: replaces the algebraic tensor product interpretation of ⊗ (Hilbert spaces, SVD, coupling matrices, ket notation) with the Structural Grammar formalism (Lambek calculus, monoidal product, categorical functors). Covers 3 phases: M3 surgical edits, M2 property renaming (automated script), M1/M0 verification. Use this skill whenever Michel mentions the structural grammar reform, the migration of hasTensorFormula, the Lambek calculus foundation, removing Hilbert space references from M3, or renaming tensor properties in M2. Also use when asked to apply the Structural Grammar reframing to any ontology file in the TSCG framework. Prerequisite: read references/Structural_Grammar_Foundation.md before starting any phase.

Orchestrates Human, LLM, and Formal Tools for safe ontology engineering through a 6-phase validation pipeline. Solves the Ouroboros paradox (LLMs need oversight but lack ontological intuition) with explicit role boundaries: Human for semantic architecture, LLM for technical execution, Formal Tools for logical validation. Covers initialization, pre-diagnosis, change execution, technical validation (linter/reasoner/SHACL), semantic validation, and documentation. Use this skill EVERY TIME you modify TSCG ontologies to ensure systematic validation and prevent semantic drift. Includes decision trees, validation checklists, and concrete scenarios (property renaming, architectural changes, notation cleanup, migrations).

Generate SHACL grammar schemas for TSCG meta-levels M1, M2, and M3. Each level has distinct validation requirements based on its architectural role. Use this skill when Michel mentions "grammaire SHACL pour M1/M2/M3", wants to validate M1 extensions (Biology, Physics, etc.), needs to validate M2 GenericConcepts structure, or must update M3 after foundational changes (e.g., structural grammar migration). The skill presents interactive level selection (M1/M2/M3) unless explicitly specified, then generates comprehensive SHACL constraints with absolute URLs, forbidden patterns, and detailed validation messages.

Create interactive 3D simulations for TSCG instances (Poclets, Systemic Frameworks, Symbolic System Grammars, etc.) using BabylonJS with modular architecture. This skill generates browser-based 3D simulations following standardized UI layout with ESLint-validated code. Works in 4 sequential steps: Inputs & Layout → Visualization Conceptualization → Prototype Creation & Iteration → Manual Testing & Finalization. Use when user wants to create a 3D simulation, mentions "visualization", "BabylonJS", "interactive simulation", or asks to visualize a TSCG instance. Prerequisites: Instance must be modeled using tscg-instance-pipeline skill first (M0_*.jsonld and M0_*_README.md required). Do NOT use for Case Studies or systems not yet modeled in TSCG.

Complete pipeline for creating a TSCG Instance in 4 sequential steps: Proposition, Analysis, Modeling, and Simulation. Use this skill whenever the user mentions an instance, wants to model a system with TSCG, wants to create an M0 JSON-LD ontology, or proposes a system/process to analyze through the TSCG lens (ASFID/REVOI dimensions, M2 GenericConcepts, M1 extensions). Also applicable when the user asks "what do you think?" about a known system, or requests an HTML/Electron simulation of a TSCG system. Do NOT use for Case Studies or Real World Systems (too complex) — this pipeline is reserved for Instances.

Pipeline de rédaction ou mise à jour d'un article de recherche TSCG en 5 étapes séquentielles : Audit, Planification, Rédaction section par section, Révision globale, Finalisation. Utilise ce skill dès que Michel mentionne l'article de recherche, demande de rédiger ou mettre à jour un article sur TSCG, veut préparer une soumission HAL ou Zenodo, mentionne "matières premières pour l'article", ou dit des choses comme "on met à jour l'article", "nouvelle version de l'article", "soumission HAL", "il faut écrire la section X", "l'article est obsolète", "version 4 de l'article". Applicable aussi quand Michel fournit un article existant à réviser ou demande d'exploiter les fichiers dans docs/papers/inputs/ ou ontology/docs/.