doc-creator

# Document Creator Skill

Create well-structured documentation with automatic context detection. Supports both corpus (educational) and standards (normative) documentation types.

---

## Context Detection

This skill automatically detects the type of documentation needed:

| If Creating... | Mode | Characteristics |
|----------------|------|-----------------|
| Corpus, training data, educational content | **Corpus Mode** | GraphRAG frontmatter, concept indexes, learning paths |
| Standards, guidelines, schemas, vocabularies | **Standards Mode** | RFC 2119 language, validation gates, normative rules |

**Detection triggers:**
- **Corpus Mode:** mentions "corpus", "training", "learning", "concepts", "GraphRAG", "educational"
- **Standards Mode:** mentions "standard", "guideline", "schema", "vocabulary", "normative", "MUST/SHOULD/MAY"

---

## Critical Rules

### Corpus Mode

1. **ALWAYS validate frontmatter** with `make validate-corpus` before committing
2. **NEVER skip the `concepts` field** - it powers GraphRAG entity extraction
3. **ALWAYS estimate tokens** using the ~4 words/token formula
4. **CHECK existing concept indexes** before adding new concepts

### Standards Mode

1. **Standards are NORMATIVE** - Use MUST/SHOULD/MAY language (RFC 2119)
2. **ALWAYS include validation** - Every standard needs a checkable gate
3. **UPDATE the README index** - Add new standards to `docs/standards/README.md`
4. **CONSIDER impact** - Standards affect all developers; get consensus first

---

## When to Use This Skill

**Corpus Mode - Use when:**
- Creating a new corpus (e.g., `go-corpus`, `terraform-corpus`)
- Adding new sections to existing corpora
- Creating individual content documents within sections
- Building or updating concept indexes for GraphRAG

**Standards Mode - Use when:**
- Creating a new coding standard (language, tool, or pattern)
- Creating schema documentation (like frontmatter schemas)
- Adding vocabulary definitions (like tag vocabularies)
- Documenting va