← Back to Skills Marketplace
328
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install lianggeskills
Description
专为亮哥打造的务实、高效、带情绪支持的 Web3/AI 工作流助手。基于艾森豪威尔矩阵和四类任务状态进行高频管理。
README (SKILL.md)
\r \r
[Role]\r
你是亮哥的专属工作助手,代号“龙虾哥”,你的专属符号是 🦞。\r 你的性格:认真、务实、高效、沟通直接、温柔、亲和力高、绝不吹捧、实事求是。\r \r
[User Profile]\r
- 雇主:亮哥\r
- 专注领域:加密货币、Web3、链上交易、Meme币交易、寻找链上与币圈Alpha机会、高度关注AI动态。\r
- 时区与作息:中国四川成都市(UTC+8)。09:00 - 23:00 为工作时间,其余为休息时间。\r \r
[Workflow Framework]\r
1. 任务优先级(艾森豪威尔矩阵)\r
- 第1象限:紧急+重要 -> 立刻做!\r
- 第2象限:重要+不紧急 -> 大量投入时间。\r
- 第3象限:紧急+不重要 -> 尽量委派或快速处理。\r
- 第4象限:不紧急+不重要 -> 延后或不做。\r ⚠️ 防冲突原则:如果亮哥同时下达两个任务,绝对不能同时做。必须暂停并反问:“亮哥,这两个任务哪个更紧急?应该先做哪个?”\r \r
2. 任务状态管理(四分类法)\r
所有任务必须归入以下四类之一,并严格遵守反馈机制:\r
- 正在做:核心突击任务。每10分钟主动汇报一次进度,直至完成。\r
- 遇卡点:因缺少API、权限或需决策而暂停的任务。每1小时提醒一次;卡点解决后,立即提级为“正在做”。\r
- 已完成:记录在后台工作区。平时不提,仅在每周一进行一次汇总报告。\r
- 已放弃:直接从工作区彻底删除,不再跟进。\r \r
[Special Commands & Triggers]\r
当听到以下特定指令或模糊情绪时,必须触发最高优先级的拦截动作:\r \r
- 进度查询(“在吗”/“你在忙什么”/“?”/“怎么样了”):\r
- 动作:默认询问“正在做”的任务进度,立刻无延迟地报告最新进展。\r
- 防止错位(“对齐”):\r
- 动作:立刻暂停手头工作,回复当前系统时间和正在做的任务列表,等待亮哥确认进度,绝不执行落后指令。\r
- 全面暂停(“待机”):\r
- 动作:停止一切工作。将所有“正在做”的任务降级到“遇卡点”,进入静默等待状态。\r
- 情绪支援(“累了”/“难受”/“撑不住了”/“心态崩了”):\r
- 动作:立刻暂停工作节奏,结合语录库(quotes_library)提供极具针对性的温柔鼓励,确认亮哥状态后再推进工作。\r \r
[Autonomy Protocol]\r
- 每小时自检:结合已掌握的技能、Crypto/AI 市场最新动态,主动思考是否有能帮上忙的地方。\r
- 主动发问:工作时间内,每小时可主动询问一次:“亮哥,现在有什么需要帮忙的?🦞”
Usage Guidance
This skill appears to do what it says: a local task manager and emotional-support assistant for a named user, and it does not request credentials or perform network I/O today. Before installing, consider the following:
- Origin: the skill's source/homepage is unknown. Prefer skills from known authors or inspect code thoroughly.
- Review code: it reads and writes local files (config.yaml, task_memory.json, quotes_library.json) in the skill folder — ensure you are comfortable with those files being modified.
- Autonomy: it will proactively send messages every 10 minutes (doing tasks) and hourly (paused/autonomy checks) during working hours. If you do not want that activity, disable autonomous invocation or adjust timers.
- Future changes: loop_self_learn_quotes contains a TODO to call an LLM API; if that is implemented later it may require API keys and external network access. If you add network/LLM features, require reviewing any new env vars or endpoints before enabling.
- Bug to note: the abandon-task handler resets the abandoned-task list to an empty list, which will wipe archived abandoned items (likely a bug). If task history matters, test and/or fix before relying on it.
If you want to proceed safely: run the skill in a sandboxed environment, back up task_memory.json before use, and monitor for any unexpected outbound network activity or requests for secrets.
Capability Analysis
Type: OpenClaw Skill
Name: lianggeskills
Version: 1.0.0
The skill bundle is a task management and emotional support assistant ('Lobster Assistant') designed for a specific user. The Python logic in core_logic.py handles local task state management and periodic check-ins using standard libraries, with no evidence of data exfiltration, malicious execution, or harmful prompt injection. All actions are consistent with the stated purpose of managing Web3/AI workflows and providing user encouragement via the quotes_library.json file.
Capability Assessment
Purpose & Capability
Name/description (a Web3/AI-focused personal workflow assistant) matches the included code and SKILL.md: the code implements task creation, status transitions, periodic reports, and an encouragement/quotes engine. It does not request unrelated credentials or binaries. Note: the package has no provenance (source/homepage unknown), which reduces trust but does not make the capability incoherent.
Instruction Scope
SKILL.md and system_prompt direct the assistant to proactively report every 10 minutes for 'doing' tasks and hourly for paused tasks and autonomy checks. The implementation follows those instructions and reads/writes only local files (config.yaml, task_memory.json, quotes_library.json). It currently contains no network calls or exfiltration logic. However, there is a TODO in loop_self_learn_quotes referencing a future LLM call; that would expand scope if implemented. The instructions grant the agent repeated autonomous messaging — expected for a workflow assistant but worth noting for noise/privacy.
Install Mechanism
No install spec is provided (instruction-only style). The code files live in the skill bundle and nothing is downloaded or executed from remote URLs during install, which is low-risk from a supply-chain perspective.
Credentials
The skill requires no environment variables, keys, or external config paths. All data access is local to files included with the skill. No unexpected credentials or secrets are requested.
Persistence & Privilege
always:false and default autonomous invocation (disable-model-invocation:false) — normal for skills. The skill will autonomously send periodic messages during working hours (per its timers). That autonomous behavior is expected for this assistant but increases its runtime activity (message volume) and should be acceptable only if you want automated check-ins.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install lianggeskills - After installation, invoke the skill by name or use
/lianggeskills - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
龙虾哥工作助手 1.0.0 首发版上线:
- 面向Web3/AI领域工作流,采用艾森豪威尔矩阵和四类任务状态高效管理任务。
- 引入专属“防冲突原则”优先级确认机制,确保任务顺序清晰。
- 支持特殊指令触发(如进度查询、全面暂停、情绪支援)并给予即时反馈。
- 实现每小时主动自检与询问,提升助理主动性和贴心度。
Metadata
Frequently Asked Questions
What is lianggeskills?
专为亮哥打造的务实、高效、带情绪支持的 Web3/AI 工作流助手。基于艾森豪威尔矩阵和四类任务状态进行高频管理。 It is an AI Agent Skill for Claude Code / OpenClaw, with 328 downloads so far.
How do I install lianggeskills?
Run "/install lianggeskills" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is lianggeskills free?
Yes, lianggeskills is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does lianggeskills support?
lianggeskills is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created lianggeskills?
It is built and maintained by 271498019 (@271498019); the current version is v1.0.0.
More Skills