Feature: Skill Graph

Skill Graph (F08)

knowledge/INDEX_DOUTO.md — The root node of Douto’s knowledge base. It maps 8 legal domains into a navigable hierarchy following the Obsidian knowledge management pattern: INDEX (root) leads to MOCs (Maps of Content) which lead to individual book entries, which will eventually lead to atomic notes.

Overview

The skill graph serves two purposes:

1. Human navigation — A lawyer or researcher can start at INDEX_DOUTO.md, click through to a domain MOC, find a relevant book, and drill down to the specific chapter or legal concept they need.
2. Programmatic access — Each file has YAML frontmatter with structured metadata (type, domain, description), enabling scripts and agents to traverse the knowledge base, discover content, and build aggregations.

graph TD
    INDEX["INDEX_DOUTO.md<br/>(root node)"]

    CIVIL["MOC_CIVIL.md<br/>35 books, ~9,365 chunks"]
    PROC["MOC_PROCESSUAL.md<br/>8 books, ~22,182 chunks"]
    EMP["MOC_EMPRESARIAL.md<br/>7 books"]
    CONS["MOC_CONSUMIDOR.md<br/>placeholder"]

    TRIB["MOC_TRIBUTARIO.md<br/>NOT YET CREATED"]
    CONST["MOC_CONSTITUCIONAL.md<br/>NOT YET CREATED"]
    COMP["MOC_COMPLIANCE.md<br/>NOT YET CREATED"]
    SUC["MOC_SUCESSOES.md<br/>NOT YET CREATED"]

    INDEX --> CIVIL
    INDEX --> PROC
    INDEX --> EMP
    INDEX --> CONS
    INDEX --> TRIB
    INDEX --> CONST
    INDEX --> COMP
    INDEX --> SUC

    style CIVIL fill:#2d5016,color:#fff
    style PROC fill:#2d5016,color:#fff
    style EMP fill:#2d5016,color:#fff
    style CONS fill:#7a6c00,color:#fff
    style TRIB fill:#8b0000,color:#fff
    style CONST fill:#8b0000,color:#fff
    style COMP fill:#8b0000,color:#fff
    style SUC fill:#8b0000,color:#fff

The 8 Domains

#	Domain	MOC File	Status	Books	Chunks
1	Direito Civil	MOC_CIVIL.md	Active	35	~9,365
2	Direito Processual Civil	MOC_PROCESSUAL.md	Active	8	~22,182
3	Direito Empresarial	MOC_EMPRESARIAL.md	Active	7	—
4	Direito do Consumidor	MOC_CONSUMIDOR.md	Placeholder	0	0
5	Direito Tributario	—	Missing	—	—
6	Direito Constitucional	—	Missing	—	—
7	Compliance & Governanca	—	Missing	—	—
8	Sucessoes & Planejamento Patrimonial	—	Missing	—	—

Summary: 4 MOC files exist (3 active, 1 placeholder). 4 MOC files are missing. Creating them is tracked as F25 (P1, milestone v0.3).

INDEX_DOUTO.md Structure

The index file uses a specific format defined in CLAUDE.md:

Frontmatter

---
type: skill-graph-index
description: Mapa do conhecimento juridico doutrinario da FS.
              Ponto de entrada para navegacao tematica.
domains: [civil, consumidor, empresarial, tributario,
          processual, constitucional, compliance, sucessorio]
---

Content

Each domain is listed under a heading with a wikilink to its MOC and a brief description of coverage:

### Direito Civil
[[MOC_CIVIL]] — obrigacoes, contratos, responsabilidade civil, propriedade

### Direito do Consumidor
[[MOC_CONSUMIDOR]] — relacao de consumo, responsabilidade objetiva, praticas abusivas

Cross-Domain Links

The index also references cross-cutting concepts that span multiple domains:

## Cross-Domain
- [[teoria_geral_do_direito]] — conceitos que cruzam areas
- [[principios_constitucionais]] — base para todos os ramos

Note

The cross-domain notes (teoria_geral_do_direito, principios_constitucionais) are referenced via wikilinks but do not yet exist as files. They are part of the planned atomic notes layer (F21/F36).

Navigation Pattern

Douto’s knowledge base follows three conventions from CLAUDE.md:

1. Wikilinks for Internal Navigation

All internal references use Obsidian-style wikilinks ([[target]]), never markdown relative links. This ensures compatibility with Obsidian’s graph view and link resolution, and makes it trivial for scripts to parse and traverse connections.

# In INDEX_DOUTO.md:
[[MOC_CIVIL]] — obrigacoes, contratos...

# In MOC_CIVIL.md:
[[contratos-orlando-gomes]] — Orlando Gomes, tratado classico

# In a chunk file:
[[MOC_CIVIL]] (back-link to parent MOC)

2. Frontmatter for Programmatic Access

Every knowledge base file has YAML frontmatter with structured metadata:

File Type	Required Fields
INDEX	type: skill-graph-index, description, domains[]
MOC	type: moc, domain, description, key_authors[], total_obras, total_chunks
Book entry	tipo: livro_chunk, titulo, livro_titulo, autor, area_direito[], status_enriquecimento
Atomic note (planned)	tipo: nota, instituto, sub_instituto, ramo, fontes_normativas[], autores[], chunks_origem[]

3. Hierarchical Traversal

The knowledge base forms a strict tree:

INDEX_DOUTO.md          (1 file — root)
  |
  +-- mocs/MOC_*.md     (8 files — one per domain)
  |     |
  |     +-- book entries  (per-book directories in staging)
  |           |
  |           +-- chunks   (individual chapter .md files)
  |
  +-- nodes/            (planned — one file per legal concept)

This hierarchy enables both top-down navigation (domain exploration) and bottom-up aggregation (collecting all chunks for a given legal concept across books).

Relationship to Other Features

Feature	Relationship
MOCs (F09-F11, F19)	MOCs are the second layer of the skill graph, directly linked from INDEX
Atomic Notes (F21, F36)	Planned third layer — one note per legal concept
F25 (Create missing MOCs)	Completes the skill graph by filling in 4 missing domains
F43 (Doctrine Synthesis Engine)	Would generate Doctrine Briefs that feed into atomic notes

Planned Improvements

Planned Feature — The following improvements are on the roadmap but not yet implemented.

• F25: Create 4 missing MOCs (P1, v0.3) — MOC_TRIBUTARIO, MOC_CONSTITUCIONAL, MOC_COMPLIANCE, MOC_SUCESSOES. Even as placeholders with minimal structure, their existence completes the navigational map.
• F36: Atomic notes from enriched chunks (P2, v0.5) — Generate one note per unique instituto from enrichment metadata, creating the third layer of the hierarchy.
• F21: Complete knowledge nodes (In Progress, v0.5) — The nodes/ directory exists with only a .gitkeep. Populating it depends on decision D03 (auto-generate vs. curate vs. hybrid).