addon-domain-semantic-adaptation（附加域语义适应）

Add-on: Domain Semantic Adaptation

Use this skill when generic scaffold outputs should reflect the user's domain language (terminology, labels, workflows, and intent phrasing) without changing core architecture patterns.

Compatibility

• Works with all architect-*, addon-*, and ui-* skills.
• Recommended default in production-default mode.

Inputs

Collect:

• DOMAIN_BRIEF: one-paragraph business/domain context.
• CANONICAL_TO_DOMAIN_MAP: key term map (for example document->clause packet, user->counsel).
• PREFERRED_TONE: neutral | formal | technical | friendly.
• FORBIDDEN_TERMS: terms to avoid in UI/docs.
• PUBLIC_COPY_SCOPE: ui-only | ui+docs | ui+docs+api-names.

Integration Workflow

1. Add semantic contract artifacts:

docs/DOMAIN_LANGUAGE.md
docs/DOMAIN_DECISIONS.md

• DOMAIN_LANGUAGE.md should contain:
• domain brief
• canonical-to-domain table
• forbidden terms
• naming rules

1. Apply semantic mapping in generated outputs:

• UI labels, helper text, empty/loading/error states
• README and operational docs
• endpoint and command aliases (when PUBLIC_COPY_SCOPE allows it)

1. Preserve technical stability:

• Keep internal architecture and safety controls unchanged.
• Prefer additive aliases over breaking renames for API/CLI contracts.

1. Add validation notes:

• Record mapped terms and intentional non-mapped technical terms in DOMAIN_DECISIONS.md.

Required Template

docs/DOMAIN_LANGUAGE.md

# Domain Language Contract

## Domain Brief
<short context>

## Canonical To Domain Terms
| Canonical | Domain |
|---|---|
| user | counsel |
| document | clause packet |

## Forbidden Terms
- foo
- bar

## Naming Rules
- UI and docs must use domain terms.
- Internal safety/security identifiers may remain canonical.

Guardrails

• Documentation contract for generated code:

- Python: write module docstrings and docstrings for public classes, methods, and functions. - Next.js/TypeScript: write JSDoc for exported components, hooks, utilities, and route handlers. - Add concise rationale comments only for non-obvious logic, invariants, or safety constraints. - Apply this contract even when using template snippets below; expand templates as needed.

• Do not change security-critical semantics for stylistic reasons.
• Do not rewrite internal identifiers when that would break compatibility.
• Keep naming consistent across UI, docs, and prompts.
• Surface ambiguous term mappings as explicit review items.

Validation Checklist

• Confirm generated code includes required docstrings/JSDoc and rationale comments for non-obvious logic.

test -f docs/DOMAIN_LANGUAGE.md
test -f docs/DOMAIN_DECISIONS.md
rg -n "Canonical To Domain Terms|Forbidden Terms|Naming Rules" docs/DOMAIN_LANGUAGE.md

Manual checks:

• Generated UI/docs reflect domain terms from user prompt.
• No forbidden terms appear in user-facing copy.

Decision Justification Rule

• Every non-trivial decision must include a concrete justification.
• Capture the alternatives considered and why they were rejected.
• State tradeoffs and residual risks for the chosen option.
• If justification is missing, treat the task as incomplete and surface it as a blocker.