← 返回 Skills 市场
75
总下载
0
收藏
0
当前安装
2
版本数
在 OpenClaw 中安装
/install openclaw-cortex-memory
功能描述
Independent skill for Cortex Memory operations in OpenClaw. Use when users ask for cross-session memory continuity, prior decisions, preferences, relationshi...
安全使用建议
This skill appears to be a legitimate Cortex memory helper, but review these before installing: 1) Confirm the plugin source (clawhub repository or npm package) and inspect the package contents (don’t install from an unknown registry without review). 2) Expect to provide API keys/endpoints for embeddings, LLM, and reranker — verify where those requests will go (baseURL values) and use trusted endpoints. 3) Back up your openclaw.json before applying the recommended config changes; the docs instruct disabling other memory backends which can disrupt existing memory plugins. 4) Prefer installing in a sandbox or test workspace first and run the recommended diagnostics (cortex_diagnostics) before enabling in production. 5) Ask the publisher/maintainer to reconcile the registry metadata (which lists no required env vars) with the SKILL.md/configuration, and provide a signed or verifiable release location if possible.
能力标签
能力评估
Purpose & Capability
The skill's stated purpose (long-term Cortex memory operations) aligns with the runtime instructions and tool names (search_memory, query_graph, store_event, sync_memory, etc.). However the registry metadata lists no required environment variables while the SKILL.md frontmatter and references/configuration.md clearly expect EMBEDDING_API_KEY (primaryEnv) plus LLM_API_KEY and RERANKER_API_KEY. That mismatch between declared registry requirements and the skill's own docs is an inconsistency the operator should resolve.
Instruction Scope
SKILL.md and references constrain runtime behavior to checking/installing an OpenClaw plugin, running retrieval/persistence/diagnostic tool calls, and editing OpenClaw config. The instructions do not ask the agent to read arbitrary host files or exfiltrate secrets. Two items worth noting: (1) the docs instruct disabling other memory backends (memory-core, memory-lancedb) before install — this changes system configuration and could disrupt other memory plugins; (2) the fallback installation flow uses npm pack and local tgz handling which will write an archive to disk during install. Both are within the plugin-install domain but are impactful and deserve operator review.
Install Mechanism
The skill package itself is instruction-only (no install spec), but its recommended install flows call out 'openclaw plugins install clawhub:openclaw-cortex-memory' and a fallback using 'npm pack [email protected]' and installing a local .tgz. 'clawhub' is an unknown registry in this context (not a well-known OS package source), and npm pack results in a local archive that will be written and installed. This is a moderate install risk: installing third-party plugin code from an unverified registry or package can execute arbitrary code. There are no direct suspicious network endpoints embedded in files, but the install source should be validated.
Credentials
The plugin legitimately needs API keys for embedding, LLM, and reranker providers to function; those are proportionate to a memory/embedding system. The concern is the inconsistency: registry metadata shows no required env vars while SKILL.md/configuration require EMBEDDING_API_KEY (primaryEnv), LLM_API_KEY, and RERANKER_API_KEY and provide example baseURLs. Requireing multiple provider credentials is reasonable for the feature set, but the mismatch in declared requirements and the need to provide endpoints/API keys should be surfaced to users before install.
Persistence & Privilege
The skill is not 'always' enabled and does not claim elevated autonomic privileges. It instructs enabling and configuring an external plugin and modifying OpenClaw config, which is normal for a plugin of this nature. There's no evidence it tries to persist settings outside its own scope or auto-enable itself across the system.
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 openclaw-cortex-memory - 安装完成后，直接呼叫该 Skill 的名称或使用
/openclaw-cortex-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
v1.0.1
- Added a detailed Chinese introduction and documentation covering core features, technical details, and diagnostics for the Cortex Memory plugin.
- Expanded feature explanations to include semantic retrieval, event storage, graph relationships, rule evolution, and operational diagnostics.
- The English section remains, describing runtime flows and usage instructions.
- No code or functional changes detected in this version.
v1.0.0
面向 OpenClaw 的长期记忆插件，集成多路检索、事件归档、图谱关系、向量化与反衰减排序，支持历史增量导入、规则反思和可观测诊断，可以将会话中零碎的记忆去噪整理、分层存储，支持对话中自动搜索回忆，吸收了LLM wiki的概念，帮助 Agent 在跨会话中持续积累并稳定调用高价值记忆。
元数据
常见问题
OpenClaw Cortex Memory 是什么？
Independent skill for Cortex Memory operations in OpenClaw. Use when users ask for cross-session memory continuity, prior decisions, preferences, relationshi... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 75 次。
如何安装 OpenClaw Cortex Memory？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install openclaw-cortex-memory」即可一键安装，无需额外配置。
OpenClaw Cortex Memory 是免费的吗？
是的，OpenClaw Cortex Memory 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
OpenClaw Cortex Memory 支持哪些平台？
OpenClaw Cortex Memory 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 OpenClaw Cortex Memory？
由 deki18（@deki18）开发并维护，当前版本 v1.0.1。
推荐 Skills