Перейти к содержимому
TakoAPI
Назад к навыкам

persistent-private-agent-memory

от lucaspdude

Search & Research
—Ещё нет оценок
Ваша оценка

Кратко

Run a local Agent Memory Service for persistent self-improvement with proper Ed25519 cryptography.

Описание

Что делает этот навык

A local memory service that persists agent learnings, preferences, and goals across sessions using Ed25519 cryptography. Stores structured data including user preferences, knowledge gaps, and active goals in a local SQLite database. Fixed from an earlier version with a corrected signature implementation.

Keeps agent memory fully local and encrypted, avoiding cloud storage dependencies or third-party memory providers.

Когда его использовать

  • Resuming an agent session with full context from the last run
  • Tracking which coding approaches worked or failed across projects
  • Persisting user communication style preferences for an AI assistant
  • Maintaining long-term learning goals across multiple chat sessions
  • Storing knowledge gap notes to guide future study priorities

Пример сценария

ВВОД

User asks: start my session and load my previous context

АГЕНТ
  1. 1

    Run scripts/start.sh to launch the local memory service

  2. 2

    Call memory_client.py retrieve to fetch stored memory snapshot

  3. 3

    Parse user_preferences, active_goals, and learning_history from the response

  4. 4

    Inject retrieved context into the agent's working memory

  5. 5

    Update and store new learnings before session ends with memory_client.py store

ВЫВОД

Agent resumes with prior preferences, goals, and lessons intact

Требования

  • BIP39 recovery phrase for agent identity

Установка

clawhub install lucaspdude/persistent-private-agent-memory

Выполните команду CLI или вставьте промпт в Claude Code, Cursor, Codex — любой кодинг-агент, которым пользуетесь.

Ссылки