龙虾知识体系管理
v1.0.0基于 Markdown 文件的 AI Agent 知识管理体系。建立和维护：SOUL（人格）/ USER（用户）/ AGENTS（工作规范）/ TOOLS（工具配置）/ MEMORY（长期记忆）/ memory（每日记录）/ .learnings（反思改进）/ HEARTBEAT（定期任务）。包含多 Agent...
⭐ 0· 115·0 current·0 all-time
by吴俊@wujun2081
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
The name and description (Markdown-based AI agent knowledge system) match the instructions: creating workspace/, SOUL.md, USER.md, AGENTS.md, memory/, .learnings/, HEARTBEAT.md and rules for using them. There are no unexpected environment variables, binaries, or installs requested.
Instruction Scope
The SKILL.md explicitly instructs the agent to read and write files under a workspace directory and to run simple shell commands (mkdir). It also prescribes an autonomous HEARTBEAT task flow and broad agent lifecycle steps (STOP→SEARCH→ACT). The instructions do not request credentials or external endpoints, but they give the agent operational discretion (e.g., 'ACT' and periodic HEARTBEAT) which could lead to actions beyond simple file bookkeeping depending on what runtime tools the agent has available.
Install Mechanism
No install spec and no code files — the skill is instruction-only, so nothing is downloaded or written by an installer. This minimizes install-time risk.
Credentials
The skill declares no required environment variables or credentials. The documentation explicitly advises not to store credentials in MEMORY.md. A minor note: TOOLS.md suggests recording API addresses, SSH/file paths, and installed-skill lists — storing such information could expose sensitive data if mishandled, but the skill itself does not demand secrets.
Persistence & Privilege
always:false and no install-time persistence are appropriate. However, the HEARTBEAT periodic-task guidance plus the default ability for agents to invoke skills autonomously means the skill can be used to run recurring self-driven tasks; this is coherent with its purpose but worth reviewing relative to the agent's execution privileges (scheduler, shell, network).
Assessment
This skill is coherent with its stated purpose, but before using it verify the environment in which the agent runs: 1) Ensure the agent's runtime does not have unnecessary shell/network privileges (if it can run arbitrary commands or access the network, HEARTBEAT and 'ACT' steps could do more than file bookkeeping). 2) Never store API keys, passwords, or other secrets in the workspace files; add sensitive paths to .gitignore and follow the SKILL.md advice. 3) Review and possibly restrict HEARTBEAT/automatic scheduling — consider manual invocation until you trust its behavior. 4) Back up any important data before allowing the agent to mass-edit files. 5) If you want extra safety, run this skill in an isolated/sandboxed workspace or a container so the file writes are limited to a controlled directory.Like a lobster shell, security has layers — review code before you run it.
latestvk97fj1v9n7by5rqewdw0naysh983ez8p
License
MIT-0
Free to use, modify, and redistribute. No attribution required.