convez.ai
SkillProdutividadeAvançado
costrict-plugins-repo

LLM Wiki — Segunda Mente para Claude Code e Obsidian

costrict-plugins-repo/alirezarezvani-claude-code-skills-llm-wiki

Wiki composta e mantida por Claude — conhecimento que cresce, não é re-derivado

Ver no GitHub
InicialPluginScore · de 0 a 100

Sistema de manutenção de wiki disciplinada que integra fontes em um vault Obsidian interligado, mantendo síntese viva e cross-referenciada. Em vez de RAG descartável, o conhecimento é compilado uma única vez por fonte e evolui incrementalmente. Inclui 3 sub-agentes, 5 comandos slash, 8 ferramentas Python stdlib e templates prontos.

Instalação

$npx convez-ai costrict-plugins-repo/alirezarezvani-claude-code-skills-llm-wiki

1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.

Métricas

Estrelas
1
Forks
0
Acompanhando
0
Saúde
37%
Último commit
há 2 semanas
Versão
—
Issues abertas
0
Contribuidores
1

Casos de uso

  • Construir e manter uma segunda mente pessoal a partir de papers, artigos e documentos
  • Integrar múltiplas fontes sobre um tópico com detecção automática de contradições
  • Consultar sínteses atualizadas sem refazer análise de RAG a cada pergunta

Veredito

Ferramenta sólida e opinada para pesquisadores e construtores que querem transformar acúmulo passivo de conhecimento em um sistema ativo de compilação. Requer comprometimento com disciplina Obsidian, mas elimina re-derivação desnecessária de sínteses.

README

A tradução para o português deste README ainda está na fila.

llm-wiki

A second brain for Claude Code + Obsidian. Inspired by Andrej Karpathy's LLM Wiki gist.

Turn any LLM CLI into a disciplined wiki maintainer. You curate sources and ask questions. The LLM reads, files, cross-references, flags contradictions, and keeps a living synthesis current. Knowledge compounds instead of being re-derived by RAG on every query.

The idea in one paragraph

Most LLM+docs workflows are RAG: retrieve fragments at query time, synthesize from scratch, forget. The wiki is compounding. The LLM reads each source once and integrates it into a persistent, interlinked Obsidian vault — updating entity pages, revising concept pages, flagging contradictions, and strengthening the synthesis. The wiki is the compiled artifact; RAG is the just-in-time retrieval. This plugin gives the LLM the discipline (SKILL.md), the delegation (sub-agents), the triggers (slash commands), and the bookkeeping (Python tools) to do the job.

What's in the box

PieceWhat it does
SKILL.mdMaster skill doc — architecture, workflows, iron rules, cross-tool compat. Has context: fork so other skills can chain into it.
3 sub-agentswiki-ingestor, wiki-librarian, wiki-linter
5 slash commands/wiki-init, /wiki-ingest, /wiki-query, /wiki-lint, /wiki-log
8 Python toolsStandard library only: init_vault, ingest_source, update_index, append_log, wiki_search (BM25), lint_wiki, graph_analyzer, export_marp
8 reference docsSchema, page formats, ingest/query/lint workflows, Obsidian setup, cross-tool setup, Memex principles
Vault templatesCLAUDE.md, AGENTS.md, .cursorrules, index.md, log.md, plus 5 page templates (entity, concept, source, comparison, synthesis)
Example vaultA small worked example on "LLM interpretability"

Quick start

# 1. Initialize a vault
python scripts/init_vault.py --path ~/vaults/research --topic "LLM interpretability" --tool all

# 2. Open in Obsidian
open -a Obsidian ~/vaults/research

# 3. Drop a source into raw/ and ingest
cp ~/Downloads/paper.pdf ~/vaults/research/raw/papers/
cd ~/vaults/research
# in Claude Code:
> /wiki-ingest raw/papers/paper.pdf

# 4. Ask questions
> /wiki-query "what does the paper say about sparse features?"

# 5. Health check
> /wiki-lint

Cross-tool compatibility

The scripts are pure Python stdlib — they run anywhere. Only the schema loader changes per tool:

ToolLoader file
Claude CodeCLAUDE.md
Codex CLI (OpenAI)AGENTS.md
Cursor (modern)AGENTS.md
Cursor (legacy).cursorrules
Antigravity (Google)AGENTS.md
OpenCode / PiAGENTS.md
Gemini CLIAGENTS.md

init_vault.py --tool all installs all three. You can run multiple CLIs against the same vault.

See references/cross-tool-setup.md for per-tool instructions.

Architecture

<vault>/
├── raw/                    # IMMUTABLE sources (you own)
├── wiki/                   # LLM-owned knowledge base
│   ├── index.md            # content catalog
│   ├── log.md              # append-only timeline
│   ├── entities/           # people, orgs, places, products
│   ├── concepts/           # ideas, theories, frameworks
│   ├── sources/            # one summary per ingested source
│   ├── comparisons/        # cross-source analyses
│   └── synthesis/          # high-level overviews and theses
├── CLAUDE.md               # schema for Claude Code
├── AGENTS.md               # schema for Codex/Cursor/Antigravity
└── .cursorrules            # (optional) legacy Cursor

Iron rule: The LLM never edits raw/. All writes go to wiki/.

Three operations

  • Ingest — Read a source, discuss with user, write summary page, update 5-15 cross-referenced pages, update index, log it
  • Query — Read index first, drill into 3-10 pages, synthesize answer with inline citations, offer to file back as a new page
  • Lint — Mechanical + semantic health check; surface contradictions, orphans, stale claims, cross-reference gaps

Why not just RAG?

Plain RAGLLM Wiki
Rediscover knowledge each queryKnowledge accumulates
Cross-references re-computed every timeCross-references pre-written and maintained
Contradictions surface only if you askContradictions flagged during ingest
Exploration disappears into chat historyGood answers re-filed as new pages
Scales by embeddings infrastructureScales by markdown + index.md + optional local search

The wiki and RAG aren't opposites — RAG can sit on top of the wiki once you outgrow index-first search.

Status

v1.0.0 — initial release. SKILL + 3 agents + 5 commands + 8 scripts + 8 references + full vault templates + example vault.

License

MIT.

Related

  • Karpathy's original gist — the pattern this plugin implements
  • Vannevar Bush, "As We May Think" (1945) — the Memex
  • qmd — local hybrid search over markdown (pair with this when the wiki outgrows index.md)
Ver o original no GitHub

Bom para

Relacionados