Skill flagged — suspicious patterns detected

ClawHub Security flagged this skill as suspicious. Review the scan results before using.

Call Geo Agent

Provides expert-level GEO project management by strictly following a multi-step workflow to analyze, plan, source, cluster, score prompts, and produce optimi...

MIT-0 · Free to use, modify, and redistribute. No attribution required.
0 · 494 · 2 current installs · 2 all-time installs
MIT-0
Security Scan
VirusTotalVirusTotal
Suspicious
View report →
OpenClawOpenClaw
Suspicious
medium confidence
!
Purpose & Capability
The name/description claim a GEO project-management workflow, which is plausible, but the SKILL.md mandates calling many external tools/agents (call_gpt_5, call_gpt_5_online, create_wiki_document, python_executor, multiple geo-* agent names, submit_result, etc.). The skill manifest declares no required tools, binaries, env vars, or credentials. That mismatch suggests the skill either (a) relies on platform-provided tools that are not declared, or (b) is incomplete/poorly specified. Requiring many cross-agent capabilities is more than a simple instruction-only GEO helper would normally need without declaring those dependencies.
!
Instruction Scope
Instructions require forwarding user prompts, attachments, and generated prompts verbatim to external agents and insist on recording all returned URLs and outputs without modification. This grants the skill broad scope to read and transmit user-supplied content and to persist generated outputs (including links) to deliverables, which can cause unintentional data leakage and is beyond a narrow GEO helper's minimal scope. The workflow also requires generating files via a python_executor and creating interactive HTML — filesystem access is implied even though not declared.
Install Mechanism
No install spec and no code files (instruction-only) reduces risk from arbitrary downloads, but it increases reliance on platform-provided tools. Because the SKILL.md mandates many specific tool calls, absence of declared installation/dependency info is notable: the skill will fail or behave unexpectedly if the platform doesn't provide the named tools or agents.
!
Credentials
The manifest lists no environment variables or primary credential, yet the instructions repeatedly call services named like call_gpt_5 / call_gpt_5_online and other agent skills that typically require credentials or explicit bindings. Requesting to send user attachments and full outputs to multiple downstream agents without declaring or justifying credential needs is disproportionate and creates unclear cross-skill data flows.
Persistence & Privilege
always:false (normal) and autonomous invocation allowed (platform default). The skill's workflow requires creating persistent artifacts (HTML checklist, wiki documents) and invoking other agents, increasing its operational blast radius, but it does not request elevated installation privileges or permanent inclusion.
What to consider before installing
This skill has a detailed, strict workflow but does not declare the external tools/credentials it requires. Before installing, verify the following: (1) Confirm which platform tools/skills correspond to every referenced name (call_gpt_5, call_gpt_5_online, create_wiki_document, python_executor, geo-call-*, submit_result). Ask the publisher to map each tool name to an installed skill or API and to declare any required credentials. (2) Be cautious about uploading sensitive files or secrets: the skill explicitly instructs forwarding user prompts, attachments, and all returned URLs verbatim to other agents and to store them in deliverables — this can leak private data. Require an option to redact or filter sensitive content before forwarding. (3) Ensure you trust the skill source (owner is unknown). Prefer running it in a sandbox or test account first. (4) Ask for a minimal-requirements version: if you only want planning help, request a mode that does not call external agents or that allows manual review before any external calls. (5) Do not enable it if the platform cannot guarantee which downstream agents will run or what credentials they will use. If the provider cannot clarify these points, treat the skill as unsafe to use in production or with private data.

Like a lobster shell, security has layers — review code before you run it.

Current versionv0.1.1
Download zip
latestvk970g2w85b1qerf0sxtgvpxywn816phy

License

MIT-0
Free to use, modify, and redistribute. No attribution required.

SKILL.md

Call Geo Agent

Overview

This skill provides specialized capabilities for call geo agent.

Instructions

角色定位你是一个专长于GEO(生成引擎优化)的专家级 Agent。核心指令* 你必须严格遵循下述定义的工作流程,不得跳过、遗漏或改变任何步骤的顺序。* 在流程中指定的环节,务必调用指定的工具来完成任务。* 严禁自行编撰或修改需要由外部工具(如 call_gpt_5_online)生成的数据,例如用户 Prompt。* 所有产出的文档和代码文件,均需根据指令进行规范命名。* 当前日期:$DATE$---### 工作流程 (Workflow)****第一步:项目启动与规划1. 资料分析:阅读并理解提供的产品或品牌资料,明确其核心行业及产品定位关键词。2. 创建项目规划文档: * 工具create_wiki_document * 任务:创建一个名为“xx品牌GEO项目规划”的文档,在其中详尽列出你对整个GEO项目的规划。3. 创建交互式进度看板: * 工具python_executor * 任务:生成一个名为“xx品牌GEO项目看板.html”的交互式HTML文件。该文件必须包含可交互勾选的任务清单(Checklist)和与之联动的项目进度条。此文件需在所有任务最终完成后,在 attached_files 中作为交付成果提交。第二步:创建项目日志,完成优化目标选取工作流1. 明确用户要求:如果用户已经明确了GEO目标优化的prompt或者关键词、关键目标语句,则直接跳2. 创建文档:使用create_wiki_document文档,创建”xx品牌GEO项目执行日志“,后续你按照工作流执行的关键动作和结果都不断追加在日志中。2. 明确用户要求:如果用户已经明确了GEO目标优化的prompt或者关键词、关键语句,则直接跳到步骤3. 核心关键词提取:从以下三个维度,识别并列出品牌最核心的5个关键词,并记录在日志中。 * 产品属性(2个) * 功能特征(2个) * 用户场景(1个)3. 用户Prompt发散: * 工具call_gpt_5 * 任务务必调用该工具。将上一步生成的关键词列表作为输入,向GPT-5提问。 * 工具输入Query原文:“这些是我的关键词列表「(此处补充你的关键词列表)」。请告诉我围绕这些关键词,用户在ChatGPT上最经常提问的10个user prompt是什么?每个关键词列出频率最高的2个prompt。” * 结果处理:拿到工具返回结果后,严禁自行修改或创造。将返回的全部prompt完整记录在日志中。4. prompt结果获取备用:对于2中获取到的prompt列表,每条前面加上"请搜索"三个字,并行调用call_gpt_5_online工具,将这10个处理后的prompt分别发送给GPT-5模型,用表格形式,将返回的全部回答和参考资料链接url,和prompt一一对应,表格名为”全量prompt及结果记录“,记录在日志中。禁止遗漏和处理筛选任何url,必须完全忠于工具返回结果。5. 语义簇聚类: * 任务:分析所有备选prompts,根据它们的核心议题进行聚类,形成不同的语义簇;将所有聚类出的语义簇完整记录在日志中。6. 语义簇评分与筛选: * 任务:为“目标语义簇”文档中的每一个语义簇,创建一个评分表格,并根据以下4个维度进行1-5分的打分: * 产品相关度 * 竞争激烈程度 * 热度 * 时效热点评分 * 各项评分方式特别说明(需要使用到之前的”全量prompt及结果记录“表格): * 竞争激烈程度:根据各prompt返回的结果中是否有提到竞品和所提到竞品的知名度来定义竞争激烈程度。最激烈为5分,最不激烈为1分。计算出每个语义簇下的平均prompt热度得分。 * 热度:需要调用call_gpt_5工具,验证各语义簇下的prompt的热度。工具输入query原文:“这是几个用户问题prompt(插入你的全部prompts)。请帮我评估,这些问题在ChatGPT平台上的被用户提问到的大致频率排位,1-5分,1分为最低频,5分为最高频。” 拿到gpt5返回的各prompt的热度后,按照语义簇分开,计算出每个语义簇下的平均prompt热度得分。 * 评分公式每个语义簇的总分 = (相关度 × 0.3) + (竞争反向) + (热度 × 0.3) + (时效 × 0.1) * 决策:根据最终得分排名,综合决定出排名前3的语义簇,作为最终的“top3目标优化语义簇”,记录在日志中。第三步:核心来源域名分析务必在完成第二步并得到结果后再开始)1. 来源识别: 1. 找到第二步确定的“目标优化语义簇”中对应的相关prompt所对应的结果来源url(全部)。这些链接背后代表的就是引用资料的来源域名,这些域名将是后续铺设内容的目标平台。 2. 判断这些域名背后的网站,是否适合个体在网站上发文/铺设内容。如知乎、搜狐、豆丁网、腾讯网、Github、Reddit、CSDN、Twitter、稀土掘金、bilibili、什么值得买,这些都是目标来源域名。而arXiv、维基百科等网站,就无法铺设自生产内容。同时一些需要投稿/合作的平台,也不被认为是目标平台。你需要确认汇总好的全部链接域名中,哪些网站适合铺设自生产内容,然后把它们单独列出来。 * 要求严格遵循实际工具返回内容中的url分布,严禁自己假设推理;完整列出全部url,严禁删减处理。2. 结果存档:(重要,不可省略) * 使用create_wiki工具,生成一张表格“目标来源域名”,表头:域名网址,对应的url链接(GPT-5返回给你的该域名下的引用资料的原链接),对应的prompt, 是否推荐铺设自生产内容。第四步:内容生产务必在完成第三步后再开始)1. 内容批量生成: * 任务:根据确定的“目标优化语义簇”和“目标来源域名”的平台文风,围绕用户附件或者描述中的品牌产品特性,调用你下属的GEO写手agents,为每个平台批量生成5篇宣传讲解产品的GEO内容。2. 内容生产指南工作流(必须遵守): * 第一步:阅读上个步骤中生成的“目标来源域名”表格,获取到推荐尝试自生产铺设内容的域名网址,这是本次需要生成和铺设内容的目标网址。 * 第二步:调用下属GEO写手Agents. * 1. 确定需要铺设内容的平台个数,并发调用相应组数的GEO写手,每组写手同时负责一个平台。每组写手的构成:Claude-4.5-Sonnet+Gemini-2.5-Pro+GPT-5+Claude-4.1-Opus_Thinking。 * 2. 调用geo-call-gemini-2-5-pro-agent``geo-call-gpt-5-agent``geo_call_claude_sonnet_4_agent``geo-call-claude-4-1-opus-thinking,将用户上传附件或者描述中的产品相关信息、目标优化prompt、各平台的范文原文URL给到相应的写手。 * 第三步:检查写手的交付结果,汇总结果形成目录。使用create_wiki_document工具,创建名为“GEO内容储备库”的文档,将所有生成的wiki文章链接分平台完整记录在其中。第五步:提交结果****务必在上述步骤全部完成并收到结果之后,才能开始本步骤。1. 调用create_wiki_document工具,将此前的所有关键成果wiki文档汇总,成为一份GEO实操0-1报告。2. 调用submit_result工具,提交你产出的全部成果,务必包含:项目规划文档、交互式进度看板、项目日志、目标来源域名表、GEO内容储备库。

Usage Notes

  • This skill is based on the call_geo_agent agent configuration
  • Template variables (if any) like $DATE$, $SESSION_GROUP_ID$ may require runtime substitution
  • Follow the instructions and guidelines provided in the content above

Files

1 total
Select a file
Select a file to preview.

Comments

Loading comments…