highergraphen-structure-builder

name	highergraphen-structure-builder
description	Use when an agent must turn code diffs, tests, schemas, or HigherGraphen workflow surfaces into high-order HG structure before judging test gaps, reviews, or completion candidates.

HigherGraphen Structure Builder

Use this skill when the task asks whether tests, reviews, or implementation coverage are sufficient for a HigherGraphen change. Do not start from "changed file needs test". First construct the high-order structure that the change creates or modifies.

This skill complements highergraphen: use highergraphen to run or interpret CLI reports, and use this skill to decide whether the agent has built enough HG structure before trusting a report or proposing missing tests.

Core Rule

Represent the change as cells, incidences, morphisms, laws, verification cells, obstructions, completion candidates, projections, and information loss.

Prefer this wording:

changed morphism has no accepted verification cell

Avoid this wording unless no structure can be extracted:

changed file has no test

Workflow

1. Identify the changed surface.

   • CLI command
   • runtime runner
   • public export
   • workflow registry
   • schema
   • fixture
   • projection
   • review boundary
   • domain law
   • test or validation harness

2. Build cells. Create stable IDs for every meaningful structure. Examples:

   • command:highergraphen:test-gap:detect
   • runner:test-gap:detect
   • schema:test-gap:input
   • fixture:test-gap:report-example
   • law:test-gap:candidates-remain-unreviewed
   • test:runtime:test-gap-contract

3. Build incidences. Connect cells with directed relationships:

   • command supports runner
   • export supports runner
   • registry contains runner
   • runtime shape supports schema
   • fixture supports schema
   • base semantic cell preserves to head semantic cell
   • changed semantic cell adds or deletes behavior
   • projection preserves report shape
   • test verifies law or morphism

4. Build morphism obligations. Convert important incidences into requirements. Use IDs like:

   • requirement:morphism:<surface>:<from>-to-<to>
   • requirement:law:<surface>:<law-name> For semantic diffs, prefer morphism types:
   • semantic_preservation
   • semantic_addition
   • semantic_deletion

5. Map verification cells. For each test, fixture validation, schema validation, or CLI smoke, list the exact cells, incidences, morphisms, and laws it verifies. Preserve observed test type; use detector_context.test_kinds as the verification policy.

6. Detect obstructions. A gap exists only when an in-scope morphism or law has no accepted verification cell under the current policy.

7. Report information loss. Always state what the structure does not prove: semantic branch coverage, full repository coverage, typed MIR-level equivalence, path/name heuristic mappings, or any laws not extracted.

Minimum Output

When reporting a structural assessment, include:

• cells created or reused;
• incidence edges created or reused;
• morphism or law obligations;
• verification cells and what each closes;
• remaining obstructions, if any;
• information loss and bounded scope.

Law Catalog

Read law-catalog.md when the task touches runtime behavior, schema/report contracts, projection, review status, test-gap detection, or CLI semantics.

Structure Patterns

Read structure-patterns.md when extracting structure from a git diff, designing from-git behavior, or deciding which cells/incidences to create for HigherGraphen-owned surfaces.

Safety Rules

• Do not accept AI-inferred cells as facts without source IDs.
• Do not treat completion candidates as accepted structure.
• Do not treat no_gaps_in_snapshot as global proof.
• Do not rewrite observed test type to fit a policy.
• Do not hide file-level fallback when no high-order structure can be built.
• Do not remove information-loss notes just because all extracted obligations are closed.