← 返回 Skills 市场
SEEM
作者
RyanToleco
· GitHub ↗
· v0.1.0
· MIT-0
84
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install seem-skill
功能描述
Advanced episodic memory system for multi-turn conversations. Store and retrieve structured conversation memories with fact graph, PPR retrieval, and three r...
安全使用建议
This skill appears internally consistent for a memory/retrieval system, but take these precautions before installing:
- Verify and set the LLM/embedding base URLs to services you trust. Default config points at third-party domains (e.g., api.deepseek.com, api.siliconflow.cn); those endpoints will receive any conversation text and images.
- Treat LLM_API_KEY and MM_ENCODER_API_KEY as sensitive secrets. Do not reuse them for unrelated accounts and prefer scoped/test keys.
- Understand the skill persists memories to disk (under the skill directory). If you do not want local persistence, disable caching/persistence in the config (enable_cache) or inspect/modify save/load methods before use.
- Because the package author and homepage are unknown, review network egress, the code paths that call the LLM/embedding APIs, and any save/load code before deploying in production.
- If you need higher assurance, request provenance (author, repo, release tag) or run the skill in an isolated sandbox to observe network traffic and file writes.
Confidence is medium because the implementation is coherent but the source is unpublished/unknown and the default endpoints are third-party; verifying endpoints and provenance would increase confidence.
功能分析
Type: OpenClaw Skill
Name: seem-skill
Version: 0.1.0
The SEEM skill bundle implements a structured episodic and entity memory system using LLMs and knowledge graphs. It is classified as suspicious due to a critical security vulnerability in `core/seem_skill.py`, where the `pickle` module is used to persist and load embedding data (e.g., `memory_embeddings.pkl`). This introduces an insecure deserialization risk that could lead to Remote Code Execution (RCE) if the local data files are tampered with. Beyond this vulnerability, the code's behavior is well-aligned with its stated purpose, and no evidence of intentional malice, data exfiltration, or harmful prompt injection was found.
能力评估
Purpose & Capability
Name/description (episodic memory, retrieval, embeddings) align with what the code asks for: python, pip, an LLM API key and an embedding API key. The required env vars (LLM_API_KEY, MM_ENCODER_API_KEY) and the Python modules listed in requirements.txt (openai, numpy, networkx, rank-bm25, etc.) are appropriate for the described functionality. Note: default base_url values in the config point at third-party endpoints (e.g., api.deepseek.com, api.siliconflow.cn); these are configurable but should be verified.
Instruction Scope
The SKILL.md and scripts direct the agent to send conversation text (and optionally images) to external LLM and embedding services and to run local CLI scripts that read/write the skill's local data. The CLI falls back to reading a local config.py if env vars are not set. There are no instructions to read unrelated system files or to transmit data to unexpected endpoints beyond the configured LLM/embed base_urls, but you should assume all stored conversation data and images will be transmitted to whatever endpoints are configured.
Install Mechanism
This skill is delivered with source files and a requirements.txt but has no automated install spec. That means installing/running it will typically require pip installing the listed packages. Dependencies are common for this domain and there are no obvious remote-download-or-extract steps in the manifest. Still, because source was published with no homepage and unknown owner, install from a controlled environment and inspect dependencies before pip installing.
Credentials
The skill requests two API keys that match its needs: an LLM key (primary) and an embeddings/MM encoder key. There are no unrelated credentials requested. Caveat: these API keys (and configured base_url values) allow the skill to send all conversation and image data to the remote LLM/embed providers you supply—ensure those providers are trusted and that keys are not reused for other sensitive services.
Persistence & Privilege
The skill persists memories and related indexes to disk (save/load logic referenced and CLI utilities create a data directory). always:false (no forced global inclusion) and disable-model-invocation:false (normal: agent or model can call the skill). This persistence is expected for a memory skill, but be aware stored data lives on the agent host and will be reloaded across runs when persistence is enabled.
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 seem-skill - 安装完成后，直接呼叫该 Skill 的名称或使用
/seem-skill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
v0.1.0
SEEM v0.1.0 – First Release
- Introduces an advanced episodic memory system with support for multi-turn conversations and structured storage.
- Features three recall modes (Lite, Pro, Max) controlling memory retrieval granularity.
- Implements hybrid retrieval strategies: Dense (DPR), Hybrid RRF (dense + sparse fusion), and graph-based PPR retrieval.
- Supports structured fact extraction, knowledge graph construction with NetworkX, and dynamic memory integration.
- Offers multimodal support, memory deduplication, and a CLI for storing and recalling memories, viewing facts, and managing state.
- Centralizes configuration for LLM and embedding models with environment variable and file-based overrides.
元数据
常见问题
SEEM 是什么？
Advanced episodic memory system for multi-turn conversations. Store and retrieve structured conversation memories with fact graph, PPR retrieval, and three r... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 84 次。
如何安装 SEEM？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install seem-skill」即可一键安装，无需额外配置。
SEEM 是免费的吗？
是的，SEEM 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
SEEM 支持哪些平台？
SEEM 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 SEEM？
由 RyanToleco（@ryantoleco）开发并维护，当前版本 v0.1.0。
推荐 Skills