本文へスキップ
TakoAPI
スキル一覧に戻る

neural-memory

nhadaututtheky 作

Productivity & Tasks
5.2kダウンロード数はclawskills.shが公開している値です
5スター
1閲覧数
0件のいいねサインインするとこのスキルにいいねできます
—まだ評価がありません
あなたの評価

概要

Associative memory with spreading activation for persistent, intelligent recall.

説明

このスキルでできること

Neural Memory gives AI agents persistent, associative memory across sessions using a spreading activation graph. Memories activate related memories rather than relying on keyword search. Supports training from documents, multi-device sync, and cognitive reasoning like hypotheses and predictions.

Spreading activation retrieval surfaces contextually related memories that keyword search would miss, making recall more human-like and coherent across long-running agent workflows.

使いどころ

  • Remembering past debugging decisions across coding sessions
  • Recalling user preferences and instructions between conversations
  • Building a knowledge base from uploaded PDFs and spreadsheets
  • Tracking project context so agents resume where they left off
  • Storing discovered patterns and insights for future reasoning

ワークフロー例

入力

User asks: 'Continue working on the authentication refactor from last week'

エージェント
  1. 1

    Call nmem_recall('authentication refactor') to load related memories via spreading activation

  2. 2

    Retrieved memories surface prior decisions, errors encountered, and chosen approach

  3. 3

    Agent resumes work with full prior context intact

  4. 4

    After completing tasks, call nmem_remember('Switched to JWT because session store caused race conditions')

  5. 5

    Call nmem_auto(action='process', text='session summary') to flush and consolidate context

出力

Agent continues the refactor with awareness of prior decisions, avoiding repeated mistakes

要件

  • OpenAI API key (if using OpenAI embedding provider)
  • Gemini API key (if using Gemini embedding provider)
  • Telegram bot token and chat ID (if using Telegram backup feature)

インストール

clawhub install nhadaututtheky/neural-memory

CLIコマンドを実行するか、Claude Code、Cursor、Codexなどお使いのコーディングエージェントにプロンプトを貼り付けてください。

リンク