⬡ TPMN-PSL v0.5.2 · Public · CC-BY 4.0

Don't write prompts.
Write specifications.

The open epistemic specification language for AI reasoning. Built on GEM² — axiomatic foundation, geometric ontology, contract-driven verification. Platform-agnostic. Formally grounded. Built to make AI know what it doesn't know.

v0.5.2 Release Read the Specs
DOI: 10.5281/zenodo.18336200 npm: @gem_squared/tpmn-skill-install

Specification documents

Three documents form the complete public reference. GEM² Core is the axiomatic foundation. PSL is the full specification. The field manual shows how to apply it.

Foundation
GEM² Core v0.5.2
5 axioms → 5 lemmas → 3 theorems + Composition Bridge (Gij). The axiomatic foundation: why F := ⟨A, B, P⟩ works.
Read
Specification
TPMN-PSL v0.5.2
Full specification — philosophical foundation, geometric ontology, compilation discipline, grammar, protocol, contract archetype, SOUND test, domain extension.
Read
Field Manual
How to TPMN v0.5.2
Practitioner's guide — step-by-step recipes for applying TPMN to real work. Compilation process, worked examples, pattern library.
Read

The difference TPMN makes

Same information. One version is auditable. One isn't.

Without TPMN
This architecture scales well and will handle production load. The latency will stay under 50ms at peak traffic. The team agrees this is the right approach.
With TPMN
This architecture handled 10k RPS in staging ⊢.
Production load is estimated at 15k RPS ⊨ (Q3 traffic model).
Latency under 50ms at peak ⊬ (no production benchmark yet).
Team consensus on approach ⊨ (3 of 4 engineers).

The 5 epistemic symbols

Every non-trivial claim in a TPMN-governed output carries exactly one tag.

⊢
Grounded
Directly supported by input or verifiable fact
⊨
Inferred
Derived from grounded claims; inference chain visible
⊬
Extrapolated
Beyond evidence; basis must be explicitly stated
⊥
Unknown
Knowledge gap; stops inference chain
?
Speculative
Possible but unverified

Geometric ontology

Two vocabularies map to Kant's noumenon / phenomenon distinction. Ontological terms name what beings ARE. Topological terms name how beings APPEAR in relation.

Ontological — noumenal

point
0D — atomic element
An entity. David, gem2-lfs, the Q3 report.
line
1D — directional vector
A verb, adjective, adverb. Failing, decomposing, becoming-stable.
face
2D — bounded local complex space ℂF
An independent space defined by CONTRACT. Humans, valid functions, completed work-plans.

Topological — phenomenal

STATE
point ∈ face
Membership relation. David is human. The WP is COMPLETED.
STATUS
vector at point
Continuous, directional. The WP is becoming complete. The build is running.
SET
subset within face
Recognized category. The set of completed WPs. Valid inputs.

Three-phase checking protocol

P-phase draws the boundary. Inline executes within it. O-phase judges the edge crossing. Operational form of human-at-the-edge.

P-phase · Boundary
Pre-flight
Extract CONTRACT from prompt. Establish A_Priori_Grid. Define the bounded face before inference begins.
Inline · Sovereign
Generation
Producer operates autonomously inside F. Tag every non-trivial claim with its epistemic symbol. Sovereignty within the contract boundary.
O-phase · Edge
Post-flight
Verify output against CONTRACT. Check SPT violations. Audit the edge crossing. Gate or pass.
Prompt → [P-phase] → LLM → [Inline] → Output → [O-phase] → Trust

The CONTRACT

F is the producer's sovereign space. The contract defines its boundary. Within this boundary, execution is autonomous.

F := ⟨A, B, P⟩

A: input type · B: output type · P: constraint
ℂF = { (a, b) | a : A, b : B, P(a,b) }

Geom(F) = A + B𝓲  —  ontology, relation, and geometry share one structure
Bounded
Each F has its own ℂF, bounded by P. No shared global plane.
Sovereign
Inside F, the producer is autonomous. The contract defines the boundary, not the orchestrator.
Composable
Gij := ⟨Bi, Aj, Pij⟩ — contract-to-contract handoff via Composition Bridge.
Verifiable
SOUND(u, D, σ) tests all five compilation principles. Per-principle audit, not pass/fail.

Structural Prohibition Taxonomy

Three inference patterns that TPMN-governed output must never produce. Each is an operational instance of CTRST failure.

S → T
State → Trait
"This model is slow" (benchmark) → "This model is inherently slow" (permanent property)
L → G
Local → Global
"Works in our staging" → "Works in all production environments"
Δe → ∫de
Incremental → Mass
One data point → broad universal claim without stated evidence count

12 lifecycle skills for Claude Code

Plan, execute, verify, archive. Contract-driven workflow that compounds knowledge across sessions.

npx @gem_squared/tpmn-skill-install
Plan
Decompose work into contracted units. A → B | P. Clarity %. No ad-hoc execution.
Verify
Check results against contracts. Field coverage, type conformance, constraint satisfaction. Structural, not opinion.
Compound
Archive proven patterns. Search prior art. Extract skills from verified work. Knowledge grows, skill count stays fixed.

Start with the spec.

Read the full TPMN-PSL v0.5.2 specification. Extend it for your domain. Build on it.

v0.5.2 Release Browse Specs