← 返回 Skills 市场

Beta Agent Memory

Name: Beta Agent Memory
Author: 1477009639zw-blip

作者 1477009639zw-blip · GitHub ↗ · v1.0.0 · MIT-0

cross-platform ⚠ suspicious

131

总下载

当前安装

版本数

在 OpenClaw 中安装

/install beta-agent-memory

功能描述

Long-term memory systems for AI agents. Implements vector memory, entity tracking, conversation summarization, and persistent context across sessions.

使用说明 (SKILL.md)

Agent Memory System

Give your AI agent persistent, long-term memory across conversations and sessions.

Memory Types Implemented

Episodic Memory

Stores episodes/events from conversations:

Key facts extracted per conversation
Decisions made and context
User preferences and patterns
"Remembering" past interactions

Semantic Memory

Structured knowledge storage:

Entity definitions and relationships
Facts about the world
Domain knowledge base
Learned procedures

Procedural Memory

Agent's own capabilities:

Known skills and tools
How to use different APIs
Response patterns that worked

Architecture

User Input
    ↓
Short-term (current session context)
    ↓
Memory Retrieval → Top-k relevant memories (vector search)
    ↓
Context Injection → Combined prompt
    ↓
LLM Response
    ↓
Memory Storage → Extract new facts, update entities

Features

Vector-based storage (ChromaDB or Pinecone)
Entity extraction (spaCy NER)
Conversation summarization (every N turns)
Relevance scoring for retrieval
Forgetting/summarization of old memories

Use Cases

Personal AI assistant that remembers you
Customer support agent with context
Research agent with persistent knowledge
Trading agent with market memory
Personal CRM (remembering people and their context)

Technical Stack

ChromaDB / Pinecone (vector store)
spaCy (entity extraction)
LangChain (memory abstractions)
PostgreSQL (structured memory)

Pricing

Type	Context Window	Price
Basic	100K tokens	$100
Pro	1M tokens	$300
Enterprise	Unlimited	$800

Built by Beta

安全使用建议

This skill describes a memory system that would normally need service credentials (Pinecone API key, DB connection, etc.), installation steps, and explicit data-governance rules—but none are provided and the source/homepage is unknown. Before installing or enabling this skill, ask the publisher: Where is memory stored (local file/Chroma/Pinecone/Postgres)? What exact environment variables and permissions are required? How is sensitive data (PII) filtered, encrypted, and deleted? Who operates the storage (third party vs you)? Prefer a version that: lists required env vars, provides an install script or code repo, documents retention/consent policies, and supports a local-only mode (Chroma/embeddings stored locally) if you need privacy. If you handle sensitive data, avoid enabling this skill until provenance, storage, and credential details are provided or until you can review the implementation.

功能分析

Type: OpenClaw Skill Name: beta-agent-memory Version: 1.0.0 The bundle contains only metadata and documentation (SKILL.md) describing a long-term memory system for AI agents. There is no executable code, suspicious commands, or prompt injection attempts present in the provided files (_meta.json and SKILL.md).

能力评估

⚠ Purpose & Capability

The description and SKILL.md consistently describe a memory system using ChromaDB, Pinecone, spaCy, LangChain, and PostgreSQL. However, the registry metadata declares no required environment variables, no install steps, and no code—yet the described capabilities normally require API keys, database connections, and package installs. That discrepancy (features that need external services but no declared credentials or installs) is incoherent.

⚠ Instruction Scope

SKILL.md tells the agent to extract and persist episodic/semantic/procedural memories, run entity extraction, periodic summarization, and use vector search. It gives no concrete runtime commands but implicitly instructs the agent to collect and store user data across sessions. There is no guidance on what to store/omit, PII handling, retention, encryption, or where/how data is persisted—this is open-ended and could lead to inappropriate or unexpected data collection.

✓ Install Mechanism

There is no install spec and no code files (instruction-only). That minimizes direct disk-write/install risk. However, the skill references Python libraries and external services that would normally require installation or service setup; the absence of an install mechanism contributes to the coherence concerns but is itself low-risk.

⚠ Credentials

The skill claims to use hosted services (Pinecone) and databases (PostgreSQL) which typically require API keys, tokens, or connection strings, yet requires.env is empty and no primary credential is declared. Requiring no credentials is disproportionate to the stated functionality and makes the specification incomplete or misleading.

ℹ Persistence & Privilege

always is false (good). The skill's purpose is persistent long-term memory; that inherently grants it privacy-sensitive persistence across sessions if implemented. The registry allows autonomous invocation (default), which combined with persistent memory behavior raises privacy concerns—especially given the lack of declared storage or governance—but autonomous invocation alone is not a disqualifying issue.

如何使用

确保已安装 OpenClaw（本地或 Docker 部署）
在对话框中输入安装命令：/install beta-agent-memory
安装完成后，直接呼叫该 Skill 的名称或使用 /beta-agent-memory 触发
根据 Skill 的参数说明提供必要输入，即可获得结构化输出

版本历史

v1.0.0

- Initial release of beta-agent-memory (v1.0.0) providing long-term memory systems for AI agents. - Implements episodic, semantic, and procedural memory types. - Features vector-based memory storage with ChromaDB or Pinecone, entity extraction via spaCy, and conversation summarization. - Supports persistent context across user sessions, enabling AI agents to recall facts, decisions, and user preferences. - Includes relevance scoring for memory retrieval and periodic forgetting/summarization of old data. - Designed for use cases like personal assistants, customer support, research agents, and CRMs.

元数据

Slug beta-agent-memory

版本 1.0.0

许可证 MIT-0

累计安装 1

当前安装数 1

历史版本数 1

常见问题