القائمة
Usar cuando se lee, escribe, busca o consolida la memoria persistente entre sesiones de Savia.
Strips all sugar-coating and marketing. Gives the brutally honest truth in the fewest possible words. Use when you suspect self-deception, before committing, or before shipping.
Adversarial review that hunts every weakness, assumption, edge case, and missing test. Opponent mode — finds what will break before it breaks in production. Use when merging, when reviewing security-critical code, or when the solution feels too simple.
Usar cuando se escribe, valida o implementa una spec ejecutable SDD.
Test-driven development with vertical-slice red-green-refactor cycles. Use when applying TDD to a new feature or bug fix, when user mentions 'red-green-refactor', 'tdd', 'test-first', 'vertical slice' — explicitly avoids the 'horizontal slicing' anti-pattern (write all tests first, then all code) which produces brittle implementation-coupled tests.
TEMPLATE — copia este directorio para crear una skill nueva. NO se carga en runtime.
Elevates perspective from trees to forest. Maps architecture, dependencies, and second-order effects before implementation decisions. Use when designing, when evaluating trade-offs, or at the start of design sessions.
| name | savia-memory |
| description | Usar cuando se lee, escribe, busca o consolida la memoria persistente entre sesiones de Savia. |
| license | MIT |
| compatibility | opencode |
| metadata | {"audience":"pm","workflow":"memory-management"} |
| trigger | {"type":"keyword","keywords":["recuerda","memory","guarda","memoriza","olvidas","recall"]} |
| consumes | ["session_data"] |
| produces | ["memory_entry"] |
Gestión de la memoria canónica externa del pm-workspace (.savia-memory/).
~/.savia-memory/
├── auto/ memoria auto (user/feedback/project/reference)
├── sessions/ snapshots de sesión
├── projects/ memoria por proyecto PM
├── agents/ memoria de agentes (public/private/projects)
├── shield-maps/ mapas mask/unmask Shield
├── pm-radar/ state.json del radar PM
└── jsonl-archive/ archivos JSONL de memoria
~/.savia-memory/auto/MEMORY.mdscripts/memory-store.shscripts/memory-store.sh search <query>scripts/memory-store.sh recall <query>scripts/memory-store.sh stats# Guardar una entrada en memoria
bash ~/claude/scripts/memory-store.sh save "<tipo>" "<contenido>"
# Buscar en memoria (search o recall)
bash ~/claude/scripts/memory-store.sh search "<query>"
bash ~/claude/scripts/memory-store.sh recall "<query>"
# Ver estadísticas de memoria
bash ~/claude/scripts/memory-store.sh stats
# Reconstruir índice desde JSONL
bash ~/claude/scripts/memory-index-rebuild.sh
~/.savia-memory/auto/MEMORY.md — índice de memoria auto.claude/profiles/active-user.md, leer preferencias y contextoauto/MEMORY.md).search (o recall) para búsqueda semántica cuando necesites contexto relacionado.Usar scripts/memory-store.sh save con el formato:
<tipo>: <descripción>
<contenido>
Tipos: decision, pattern, context, feedback, lesson, reference
❌ Guardar sin tipo: usar --type custom para todo en lugar del tipo semántico correcto (decision, discovery, bug, etc.) → memoria no recuperable por topic, búsquedas devuelven ruido.
✓ Correcto: seleccionar el tipo que mejor describe la naturaleza del dato antes de guardar.
❌ Guardar sin source: omitir --source skill:<name> o --source session → trazabilidad rota, entries huérfanas sin origen verificable.
✓ Correcto: siempre incluir --source con el skill, comando o sesión que originó la entrada.