session-memory
por swaylq
Notes & PKMResumo
Persistent memory toolkit for AI agents.
Descrição
O que esta skill faz
A set of bash and Node.js scripts that give AI agents persistent memory across sessions. Memories are stored locally as JSONL files, tagged by topic, and scored by importance and recency. Supports save, recall, consolidation, export/import, and pruning.
Runs with no external dependencies or API calls, keeping all memory local and available offline.
Quando usá-la
- Resuming a coding session by loading recent decisions and insights
- Recording an architectural choice so the agent remembers it next week
- Flagging a credential rotation as critical so it always surfaces in context
- Reviewing a week of decisions grouped by topic before a team sync
- Exporting a memory snapshot before wiping a development environment
Exemplo de fluxo de trabalho
User asks: load my context from the last 3 days and remind me what database decisions we made
- 1
Runs context.sh --days 3 to load recent and high-priority memories
- 2
Runs recall.sh "database" --topic decision to find relevant entries
- 3
Presents matched memories sorted by relevance score
- 4
Saves a new memory summarizing the current session's starting point
Agent outputs recent context, lists database decisions with dates and importance levels, and confirms the session summary was saved
Instalação
clawhub install swaylq/session-memoryExecute o comando da CLI ou cole o prompt no Claude Code, Cursor, Codex ou no agente de código que você usa.