Goblin: Systemic Closure Illusion Gradient

Repo target: nkllon/goblin

This bundle contains:

• goblin-ontology.ttl — OWL ontology defining Goblin score (SCIG), problem classes, dimensions, and example domains.
• goblin-shapes.ttl — SHACL shapes constraining Goblin score usage.
• goblin-map.dot — Graphviz DOT file for a "Goblin map" of common domains.
• goblin-agent-lim42.md — lim42-ready agent spec for computing Goblin scores.
• goblin-energy-gradient.md — energy gradient interpretation with simple flow equations.
• Rendered docs/goblin-map.svg — quick visual of the Goblin map without requiring Graphviz (generate PNG locally with dot -Tpng goblin-map.dot -o docs/goblin-map.png if you prefer).

You can pip install . (or pip install -e . for development) to get the ontology, SHACL shapes, and validation helpers as a tiny Python package (goblin-score). Then validate data against the bundled shapes (after installation):

python -m goblin.validate --data samples/goblin-sample.ttl

Meme / Usage

Because "SCIG" is overloaded in Google (subcutaneous immunoglobulin), we use the meme name:

Goblin score

Example phrases:

• "What is the goblin score for that model?"
• "That architecture looks like a goblin, what do you think?"
• "Can you compute the goblin score for this solution?"

The Goblin score is defined in the ontology as ui:goblinScore and is an alias of ui:SCIG_Score.

Next Steps

• Push this bundle into nkllon/goblin as initial commit.
• Wire goblin-agent-lim42.md into lim42 as a reusable agent profile.
• Render goblin-map.dot with Graphviz to produce a PNG/SVG for documentation. ✅

Scoring workflow (lim42 example)

Use the existing lim42 agent prompt (goblin-agent-lim42.md) and weights to go from a system description to a JSON score.

Example prompt input (shortened):

"A multi-tenant SaaS with asynchronous provisioning, eventual consistency across regions, and teams expecting deterministic dashboards. Observability is partial; on-call engineers rely on retries and manual overrides."

Expected JSON output shape (weights baked into the agent):

{
  "goblin_score": 0.82,
  "dimensions": {
    "illusion_closure": 0.90,
    "distributed_state": 0.85,
    "async_nondeterminism": 0.80,
    "stakeholder_demand_for_determinism": 0.70,
    "observability_deficit": 0.60,
    "energy_gradient_misalignment": 0.95
  },
  "commentary": "Async provisioning and partial observability create large closure illusions; teams expect deterministic dashboards."
}

Goblin map preview

• SVG — scalable version from the DOT source (goblin-map.dot).
• To produce a PNG locally (not checked into the repo), run dot -Tpng goblin-map.dot -o docs/goblin-map.png.