Task Tracker Pro

持久化任务管理：拆解步骤、追踪进度、中断恢复、多Agent协作。 触发场景： - 收到需要多步骤执行的任务（"帮我做XX"、"搞一个XX"） - 新对话/会话启动时自动检查未完成任务 - 用户问"任务进度"、"做到哪了"、"继续上次的" - 用户说"建个任务"、"拆解一下"、"规划一下步骤" 核心理念：不信任对话...

This skill appears to do what it says (persist task state to ~/.openclaw/workspace/tasks/) and asks for no credentials, which is good. However, the runtime instructions are vague about what '执行该步骤的实际工作' means — an agent following this could perform arbitrary actions (network requests, shell commands, read other files) to complete steps. Before installing or enabling it: 1) Confirm how your agent runtime restricts file and network access; run the skill in a sandbox if possible. 2) Inspect any created task files regularly and set filesystem permissions for ~/.openclaw/workspace/tasks. 3) If you want tighter control, ask for a version of the skill that explicitly limits allowed actions (e.g., only local edits to task files, no external network calls) or require explicit user confirmation before performing non-file operations. 4) Monitor logs for unexpected outbound requests. These precautions reduce the risk that vague 'do the work' instructions are used to perform unwanted operations.

Type: OpenClaw Skill Name: task-tracker-pro Version: 1.0.2 The skill is a task management utility designed to provide persistence across AI agent sessions by storing state in Markdown files. It operates within a restricted workspace directory (~/.openclaw/workspace/tasks/) and uses standard file operations (ls, mv, and file writes) to track progress. No indicators of data exfiltration, malicious execution, or harmful prompt injection were found in SKILL.md or README.md.