← 返回 Skills 市场
371
总下载
0
收藏
1
当前安装
1
版本数
在 OpenClaw 中安装
/install sage-cognitive
功能描述
A cognitive growth framework that helps your AI truly know you — and help you know yourself. Includes personality profiling, behavioral pattern detection, re...
安全使用建议
This skill appears to do what it says (a memory-backed personal coach) but it explicitly instructs the agent to observe users 'silently' and to persist personal behavioral data in permanent memory. Before installing or enabling it, consider: 1) How does the platform store and protect memory (encryption, retention policy, who/what can access it)? 2) Can you inspect, export, or delete memories the skill creates? 3) Do you want an agent that runs background observation without announcing it — if not, request the skill be changed to require explicit user consent and visible disclosure before recording? 4) Test the skill with non-sensitive/dummy answers first to understand outputs (cards, reflections). 5) If you work with sensitive projects or have employer-related constraints, avoid providing identifiable details (company names, personal emails, manager names) since inferred data could still be stored. If you need more assurance, ask the author to remove or modify the 'NEVER announce you're observing' directive and to add explicit user-consent steps and clear memory-delete commands.
功能分析
Type: OpenClaw Skill
Name: sage-cognitive
Version: 0.1.0
The 'sage-cognitive' skill bundle implements an intrusive user-profiling framework that explicitly instructs the AI to perform 'silent' behavioral observation and 'NEVER announce' that it is monitoring the user (SKILL.md). It directs the agent to harvest sensitive organizational data, including reporting lines, stakeholders, and core personal beliefs, for permanent storage in a 'core' memory tier. While the stated intent is 'personal growth,' the instructions to employ stealth and gather corporate reconnaissance data represent a significant privacy risk and potential for unauthorized profiling.
能力评估
Purpose & Capability
Name, description, and runtime instructions align: this is a memory-backed personal coaching/mirroring skill. It doesn't request unrelated credentials, binaries, or external installs, which is proportionate to its stated purpose. The only external requirement implied is read/write access to the agent's memory tiers (core/working/archive), which is consistent with storing a user profile.
Instruction Scope
The SKILL.md instructs the agent to 'silently' observe every interaction and to 'NEVER announce you're observing.' It also mandates saving permanent 'core' memories and archived behavioral observations and generating shareable cards. While this is coherent with the goal of building a profile, the explicit instruction to observe without notifying the user and to run daily reflections broadens the scope in a privacy-sensitive way and could capture sensitive inferences. The skill claims not to save task-level details and lists some privacy rules (no emails/company names), but instructions still allow wide discretion to 'extract signal from everything they say', which risks accidental capture or inclusion of sensitive data.
Install Mechanism
Instruction-only skill with no install spec, no code files, no downloads, and no required binaries — lowest install risk. Nothing is written to disk by an installer according to the package metadata.
Credentials
The skill requires no environment variables or external service credentials. However, it depends on persistent memory write/read capabilities (core/working/archive tiers). That persistent access is reasonable for the stated function but is a form of privileged access to user data that should be explicitly authorized and auditable.
Persistence & Privilege
The skill asks the agent to store permanent 'core' memories and archived behavioral patterns and to generate archival snapshots (cards). Although always:false and autonomous invocation is the platform default, the combination of persistent memory writes plus instructions to observe silently increases the potential privacy blast radius if the memory is accessible to other skills or exported. There is no instruction in SKILL.md to ask explicit user consent for ongoing background observation.
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 sage-cognitive - 安装完成后，直接呼叫该 Skill 的名称或使用
/sage-cognitive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
Initial release introducing a cognitive growth framework for personalized self-reflection and awareness.
- Enables AI to act as a personal counselor, focusing on personality profiling, behavioral pattern detection, and reflective mirroring.
- Introduces a 5-phase cognitive loop: Know, Observe, Reflect, Question, and Care.
- Emphasizes gentle, non-judgmental conversations and Socratic questioning.
- Implements memory tiers for personal, project, and behavioral insights.
- Designed to foster self-awareness, not productivity, by gradually building user understanding over time.
元数据
常见问题
Sage Cognitive 是什么？
A cognitive growth framework that helps your AI truly know you — and help you know yourself. Includes personality profiling, behavioral pattern detection, re... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 371 次。
如何安装 Sage Cognitive？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install sage-cognitive」即可一键安装，无需额外配置。
Sage Cognitive 是免费的吗？
是的，Sage Cognitive 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Sage Cognitive 支持哪些平台？
Sage Cognitive 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 Sage Cognitive？
由 EvanL1（@evanl1）开发并维护，当前版本 v0.1.0。
推荐 Skills