메뉴
Transforma a memória do Simplicio (memory_items + embeddings all-MiniLM) em um knowledge graph consultável — nós + arestas, comunidades, god nodes, path/explain/query. Port do conceito Graphify (Graphify-Labs) reusando a infra já existente (neural-recall, vec0).
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Legacy compatibility alias for `/simplicio-loop`. Use only when an older install, adapter, or saved prompt still invokes `/simplicio-tasks`; immediately route the run through `.claude/skills/simplicio-loop/SKILL.md`, which now owns the public core + loop protocol.
Operate and troubleshoot local Hermes/Simplicio gateway services safely, including config edits, launchd/service restarts, and live verification with logs.
Legacy compatibility alias for `/simplicio-loop`. Use only when an older install, adapter, or saved prompt still invokes `/simplicio-tasks`; immediately route the run through `.claude/skills/simplicio-loop/SKILL.md`, which now owns the public core + loop protocol.
Learned skill -- when the task mentions "not", run this proven sequence (5 successful runs, confidence 1.00).
Evolutionary optimize-by-metric loop — mutate a target, evaluate against fixed criteria, KEEP if the score improves (commit) or REVERT if it doesn't (git checkout), repeat, plateau-break after N stagnated runs. Adapts Karpathy's `autoresearch` pattern (and the ECC `autoresearch-agent`) with mandatory yool guardrails (§11 caps), git-isolated branch discipline, an anti-Goodhart eval order (correctness gate FIRST, score SECOND, labeled tokenizer), a local-first mutation ladder, and a `simplicio.savings-event/v1` receipt per run. Use when the user says "optimize this by <metric>", "autoresearch loop", "evolutionary optimize", "mutate/eval/keep-revert", or asks to shrink tokens/latency/bundle size/improve pass-rate against a fixed, measurable eval. Worker: `scripts/autoresearch.py`.
Cut output and memory tokens without losing meaning — terse prose levels (caveman-style) that preserve code/paths/URLs byte-for-byte, plus a one-time memory/doc compaction pass that pays back every future turn. Use when replies or worker reports are verbose, when standing context (CLAUDE.md/AGENTS.md/notes) is bloated, or when simplicio-tasks needs its output-side + input-side token discipline. Compression NEVER touches code, identifiers, or a safety confirmation.
| name | graphify |
| title | Graphify — Knowledge Graph sobre a Memória do Simplicio |
| description | Transforma a memória do Simplicio (memory_items + embeddings all-MiniLM) em um knowledge graph consultável — nós + arestas, comunidades, god nodes, path/explain/query. Port do conceito Graphify (Graphify-Labs) reusando a infra já existente (neural-recall, vec0). |
PORT do conceito Graphify (Graphify-Labs): transformar conteúdo em um
knowledge graph consultável — entidades (nós) + relações (arestas),
clusterizado em comunidades, com centralidade (god nodes) e navegação
query / path / explain.
O Simplicio Agent já tinha a metade da infraestrutura, mas faltava a camada relacional:
| Camada | Estado antes | Onde |
|---|---|---|
| Embeddings reais (all-MiniLM-L6-v2) | ✅ 34.912 vetores | memory_vectors |
| Busca híbrida FTS5+ANN | ✅ neural-recall | ~/.simplicio_agent/scripts/neural-recall |
| Tabelas de grafo no schema | ✅ existem | memory_relationships, memory_edges |
| Relações populadas | ❌ 0 arestas | vazias |
O Graphify preenche exatamente esse buraco: extrai arestas e mede o grafo.
graphify_build.py reusa os
embeddings JÁ computados e deriva arestas por:
same_repo, same_file, same_dir)def/class/import/call colocalizados no código (calls, imports)semantic_neighbor)memory_relationships (grafo vivo) + graph.json
(cache persistente) + GRAPH_REPORT.md + vault Obsidian.graphify_query.py: query (traversal + hops), path
(BFS menor caminho), explain (vizinhança do nó), report.# Build completo (determinístico, sem LLM)
python3 ~/.simplicio_agent/skills/simplicio/graphify/scripts/graphify_build.py
# Smoke test em 2000 itens (validar antes do build full)
python3 ~/.simplicio_agent/skills/simplicio/graphify/scripts/graphify_build.py --limit 2000 --dry-run
# Consultar
python3 ~/.simplicio_agent/skills/simplicio/graphify/scripts/graphify_query.py query "retry policy" --hops 2
python3 ~/.simplicio_agent/skills/simplicio/graphify/scripts/graphify_query.py path "code:x:a.py" "code:x:b.py"
python3 ~/.simplicio_agent/skills/simplicio/graphify/scripts/graphify_query.py explain "skill:simplicio-runtime:graphify"
python3 ~/.simplicio_agent/skills/simplicio/graphify/scripts/graphify_query.py report
neural-recall acha nós por similaridade; o Graphify expande o contexto
seguindo arestas do grafo (1-hop/2-hop). É o equivalente Simplicio ao
/graphify query do Graphify original — em vez de reler arquivos, navega o grafo.
--mode deep habilita inferência de arestas IMPLÍCITAS via LLM
(OpenRouter deepseek/deepseek-v4-flash). Requer OPENROUTER_API_KEY. Sem credencial,
cai automaticamente no determinístico (já cobre a maioria das relações reais).
SELECT COUNT(*) FROM memory_relationships deve subir de 0 após o build.GRAPH_REPORT.md lista god nodes e conexões surpreendentes.graph.json é consultável sem reler 35k itens.