Canonical Kernel

Base120Frozen v1.0.0

120 canonical mental-model operators organized across six transformation families: the frozen, open cognitive kernel of HUMMBL.

Base120 is the cognitive governance layer of HUMMBL. It is a corpus of 120 validated mental models used to govern how AI agents reason, not merely what they are permitted to do. The canonical registry is frozen at v1.0.0; any change requires a governed version bump.

The six transformation families

Models are grouped into six families of 20. Codes follow the pattern <FAMILY><N>, where N is 1–20.

Code	Family	Focus
P	Perspective	Frame and name what is. Anchor or shift point of view.
IN	Inversion	Reverse assumptions. Examine opposites, edges, negations.
CO	Composition	Combine parts into wholes. Build complexity from simplicity.
DE	Decomposition	Break wholes into parts. Simplify by separation.
RE	Recursion	Apply patterns across scales. Self-reference and iteration.
SY	Systems	Understand how parts interact, create emergent behavior, and shape dynamics. (Not synthesis.)

Difficulty tiers

Within each family, models are grouped into three difficulty tiers:

Tier	Range	Count per family	Total
Beginner	1–7	7	42
Intermediate	8–14	7	42
Advanced	15–20	6	36

Examples

P1 — First Principles Framing

Reduce complex problems to foundational truths that cannot be further simplified. Used when the team is optimizing inside an inherited assumption.

IN6 — Inverse/Proof by Contradiction

Assume a claim is false, derive a logical impossibility, thus proving the claim true. Used to stress-test security assumptions and architectural claims.

SY13 — Incentive Architecture

Design reward and penalty structures that align individual actions with system goals. Used when a governance change is being undermined by misaligned incentives.

DE5 — Dimensional Reduction

Focus on the most informative variables while discarding noise or redundancy. Used to simplify monitoring dashboards and incident triage.

Schema

Each Base120 model is represented as a JSON object with a stable code, definition, and metadata.

Model:
  id: str            # "P1", "SY13", "DE5"
  name: str          # "First Principles Framing"
  domain: str        # "P"
  domain_name: str   # "Perspective"
  definition: str    # Full operator definition
  difficulty: str   # "beginner" | "intermediate" | "advanced"
  keywords: list[str]

Recommendation:
  model_id: str
  score: float
  reasons: list[str] # "keyword(2.5)", "ledger(1.2)", etc.

How it is used

Base120 operators are retrieved by code, listed by family, or recommended for a natural-language problem. The recommendation engine combines keyword overlap with past usage signals from the Cognitive Ledger Protocol (CLP).

# Hosted API
GET https://api.hummbl.io/v1/models/P1

# Recommend for a problem
POST https://api.hummbl.io/v1/recommend
{"problem": "How do I prioritize features for my MVP?", "limit": 5}

Provenance

Origin: Base120 was developed as the cognitive governance substrate for HUMMBL, addressing OWASP ASI08 (Cascading Cognitive Failures) and ASI09 (Human-Agent Trust Exploitation) by giving agents named, auditable reasoning frameworks.

Canonical source: founder_mode/cognition/data/base120_registry.json and the frozen YAML registry in the Base120 repository.

Reference: BASE120-REFERENCE.md — full catalog, usage guide, and API shapes.