| name | highergraphen-ddd |
| description | Use when an agent needs to run or interpret the bounded `highergraphen ddd` review workflow for domain model, bounded context, aggregate, anti-corruption mapping, evidence-boundary, projection-loss, or closeability diagnostics. |
HigherGraphen DDD Skill
Use this skill for the repository-owned highergraphen ddd ... workflow. This
skill replaces the old CaseGraphen-specific DDD diagnostic skill for product
workflow usage: DDD interpretation belongs in the bounded HigherGraphen DDD
workflow contract, not in higher-graphen-core.
Source Of Truth
- CLI reference:
docs/cli/highergraphen.md
- DDD review contract:
docs/specs/ddd-review-cli-contract.md
- Input schema:
schemas/inputs/ddd-review.input.schema.json
- Input fixture:
schemas/inputs/ddd-review.input.example.json
- Report schema:
schemas/reports/ddd-review.report.schema.json
- Report fixture:
schemas/reports/ddd-review.report.example.json
- Legacy motivating example:
examples/casegraphen/ddd/domain-model-design/README.md
Do not restate the schemas as competing contracts. Consume the contract,
fixtures, and CLI output.
Commands
Convert a native CaseGraphen DDD case-space fixture into bounded DDD review
input:
cargo run -q -p highergraphen-cli -- \
ddd input from-case-space \
--case-space examples/casegraphen/ddd/domain-model-design/sales-billing-customer.case.space.json \
--format json \
--output ddd-review.input.json
Run DDD review on bounded input:
cargo run -q -p highergraphen-cli -- \
ddd review \
--input schemas/inputs/ddd-review.input.example.json \
--format json
Write a report:
cargo run -q -p highergraphen-cli -- \
ddd review \
--input schemas/inputs/ddd-review.input.example.json \
--format json \
--output ddd-review.report.json
Validate checked-in schema-bearing fixtures:
python3 scripts/validate-json-contracts.py
Interpretation Rules
- Follow
docs/guides/product-integration-for-ai-agents.md: start from a
bounded source snapshot, lift into a DDD review space, preserve the lift
morphism, map DDD vocabulary through explicit interpretation mappings,
evaluate invariants, then project human/AI/audit views with declared loss.
- Treat
source_boundary.id, lift_morphism, and operation_gate as required
product integration objects. The source boundary says what was read and
omitted; the lift morphism says how source records became DDD review
structure; the operation gate says which actor/capability/policy/scope is
allowed to run the diagnostic over that boundary.
- Exit code
0 means the workflow emitted a report. Domain findings such as
blocked decisions, missing accepted evidence, unreviewed completion
candidates, projection loss, review gaps, and non-closeable state are
successful report data, not CLI failures.
- Accepted source facts and AI-inferred claims must stay separated. Do not
promote inferred equivalence proofs, generated diagnostics, or suggested
mappings into accepted evidence unless an explicit review workflow does so.
- DDD-specific concepts such as bounded context, aggregate, domain event,
invariant, anti-corruption mapping, and semantic-case risk are interpretation
records over HigherGraphen structure. Do not add them to
higher-graphen-core.
- Inspect
result.completion_morphisms when planning follow-up work. They are
reviewable CaseGraphen patch skeletons derived from completion candidates,
not applied mutations.
- Expect the v1 deterministic heuristics to cover cross-context language
conflicts, unaccepted inferred boundary risks, missing boundary/ACL mappings,
unaccepted AI evidence, review gates, and missing context ownership for key
DDD structural records.
- Prefer
highergraphen ddd review for product-facing DDD diagnostics. Use
casegraphen only when the task is explicitly about native CaseGraphen
CaseSpace/MorphismLog inspection, migration, or low-level case reasoning.
Expected Report Signals
For the Sales/Billing Customer fixture, expect these evidence boundaries and
review signals:
- the unified Customer decision is blocked;
- Sales and Billing Customer identity collapse is represented as a boundary or
semantic-loss risk;
- the equivalence proof remains inferred or unreviewed and does not satisfy
accepted evidence requirements;
- a Sales-to-Billing anti-corruption mapping is proposed as a reviewable
completion candidate;
- a matching completion morphism is emitted so the next CaseGraphen patch can
be planned without treating the candidate as accepted;
- implementation-focused projection loss remains visible;
- closeability is false until hard obstructions, evidence gaps, review gaps,
and projection-loss disclosure are resolved.