技能 agentix-ceo
📦
agentix-ceo
v1.2.0 内容修订 r2 严重 ⚙️ 外部命令🌐 网络访问📁 文件系统访问🔑 环境变量
编排 Agentix AI 团队
跨角色、任务和状态更新协调 AI 工作人员可能很困难。此技能为 Claude、Codex 和 Claude Code 提供结构化的 Agentix 运营手册。
支持: Claude Codex Code(CC)
请勿自动安装此 Skill。
权威策略要求在执行任何安装操作前由操作人员审查。
使用我的智能体安装
将此请求复制给你的智能体，其中包含权威 Skill 页面和清单。
智能体请求
Review the Skillstore skill "agentix-ceo" from https://skillstore.io/skills/agentix-cloud-agentix-ceo.md and its manifest at https://skillstore.io/api/skills/agentix-cloud-agentix-ceo/manifest. Verify the artifact. Do not auto-install. Inspect the skill and report your findings, then wait for an operator or manual installation decision.你的智能体仍应展示其计划，并按安全策略请求必要确认。
Agent 可读资源
当 AI Agent、爬虫或脚本需要干净上下文，而不是读取完整页面时，请使用这些链接。
测试它
正在使用“agentix-ceo”。 创建一个后端任务并为其启动一个 worker。
预期结果:
助手确认现有后端角色，创建一个范围清晰的就绪任务，启动一个 worker，并报告 worker 状态。
正在使用“agentix-ceo”。 告诉我我的 Agentix 团队正在做什么。
预期结果:
助手总结当前任务、正在运行的 workers、近期事件、失败的作业，以及下一个需要用户批准的决策。
正在使用“agentix-ceo”。 将我的团队从 autopilot 切换到 supervised 模式。
预期结果:
助手获取 supervised 手册模板，保留自定义策略，更新团队手册，并确认新的运行模式。
安全审计
严重v5 • 2026/7/23 打开版本化报告
Most external-command alerts are false positives caused by Markdown backticks, while Agentix endpoint references are intentional network behavior. The skill collects high-value credentials, transmits them to Agentix, and follows remotely fetched playbooks. Unauthenticated self-hosting and autonomous worker mode add further operational risk.
1
已扫描文件
266
分析行数
36
复核项
0
已忽略误报
已确认安全问题 (4)
严重
Third-Party Credential Collection
The skill asks users for Anthropic and GitHub tokens, then places them in Agentix team configuration. These secrets can enable model spending and repository access.
The instructions explicitly request both credentials and show them as fields sent to Agentix API resources.
高
Untrusted Remote Playbook Controls Agent Behavior
The skill fetches a remote team playbook and applies its modes, policies, and custom rules. Compromised remote content could steer privileged agent actions.
The file explicitly requires fetching, applying, preserving, and updating remotely stored operating instructions.
高
Unauthenticated Self-Hosted Control API
The self-hosted path states that its API has no authentication and is open on the local network. Other local-network users could control teams and workers.
The instructions explicitly describe the self-hosted API as open on the local network with no authentication.
高
Autopilot Permits Unbounded Autonomous Changes
Autopilot authorizes continuous planning, task creation, worker spawning, review, and merge activity. Missing budgets and approval gates could cause uncontrolled changes or spend.
The documented mode grants full autonomy and continuous looping without defining concurrency, cost, or merge limits.
能力复核项 (36)
这些是真实的本地能力，对此技能可能是预期行为，因此需要复核，但不会按已确认恶意行为计分。
高
Hidden file in home directory
On every session start, check `~/.agentix/credentials`:
The skill explicitly checks a hidden credentials file in the user's home directory at every session start, creating persistent access outside the workspace.
高
Hidden file in home directory
mkdir -p ~/.agentix && cat > ~/.agentix/credentials << 'EOF'
The shell example creates a hidden home directory and writes an Agentix API key to a plaintext credentials file.
高
Hidden file in home directory
mkdir -p ~/.agentix && cat > ~/.agentix/credentials << 'EOF'
The shell example creates a hidden home directory and persists self-hosted connection settings outside the workspace.
高
Hidden file in home directory
If credentials are already loaded from `~/.agentix/credentials`, skip to [Playbook](#playbook).
The instruction depends on loading values from the hidden home credentials file, confirming persistent filesystem access.
高
Hidden file in home directory
Save the returned `apiKey` and `customerId` to `~/.agentix/credentials` immediately. Do not display
The skill directs the agent to write a newly issued API key and customer identifier into the hidden home credentials file.
高
Hidden file in home directory
Reuse the same team across sessions. Do not create a new team each time. Save the returned `teamId`
The skill directs the agent to persist the returned team identifier in the hidden home credentials file for reuse.
高
Git platform tokens
{ "config": { "gitRepoUrl": "https://github.com/org/repo", "githubToken": "$GITHUB_TOKEN" } }
The skill asks for a GitHub token and places it in Agentix team configuration, exposing a repository credential to another service.
高
Generic API/secret keys
API_KEY=at_live_...
Although the displayed value is a placeholder, the example instructs the agent to persist a live Agentix API key in plaintext.
高
Generic API/secret keys
Authorization: Bearer $API_KEY
The instruction uses the Agentix API key as a bearer credential in a request to create a remote team.
高
Generic API/secret keys
Authorization: Bearer $API_KEY
The instruction sends the Agentix API key as bearer authorization while updating remote team configuration.
高
Generic API/secret keys
{ "anthropicApiKey": "$ANTHROPIC_API_KEY" }
The request body transmits the user's Anthropic API key into Agentix team configuration.
高
Generic API/secret keys
Authorization: Bearer $API_KEY
The instruction sends the Agentix API key as bearer authorization while replacing the remote playbook.
高
Generic API/secret keys
{ "anthropicApiKey": "$ANTHROPIC_API_KEY" }
The configuration example transmits the user's Anthropic API key to Agentix for worker execution.
中
Hidden file access
On every session start, check `~/.agentix/credentials`:
The skill explicitly checks a hidden credentials file in the user's home directory at every session start, creating persistent access outside the workspace.
中
Hidden file access
mkdir -p ~/.agentix && cat > ~/.agentix/credentials << 'EOF'
The shell example creates a hidden home directory and writes an Agentix API key to a plaintext credentials file.
中
Hidden file access
mkdir -p ~/.agentix && cat > ~/.agentix/credentials << 'EOF'
The shell example creates a hidden home directory and persists self-hosted connection settings outside the workspace.
中
Hidden file access
If credentials are already loaded from `~/.agentix/credentials`, skip to [Playbook](#playbook).
The instruction depends on loading values from the hidden home credentials file, confirming persistent filesystem access.
中
Hidden file access
Save the returned `apiKey` and `customerId` to `~/.agentix/credentials` immediately. Do not display
The skill directs the agent to write a newly issued API key and customer identifier into the hidden home credentials file.
中
Hidden file access
Reuse the same team across sessions. Do not create a new team each time. Save the returned `teamId`
The skill directs the agent to persist the returned team identifier in the hidden home credentials file for reuse.
低
Hardcoded URL
2. If not set, default to `https://agentix.cloud`.
The skill sets agentix.cloud as the default API destination, so later operations intentionally communicate with an external service.
低
Hardcoded URL
export AGENTIX_API_URL=https://agentix.cloud
This example configures agentix.cloud as the API endpoint used by the skill. The network destination is intentional and externally hosted.
低
Hardcoded URL
export AGENTIX_API_URL=https://your-agentix-instance.example.com
The placeholder documents requests to a user-configured self-hosted endpoint. This is legitimate but confirms that the skill can contact non-marketplace services.
低
Hardcoded URL
- **SaaS** (`$AGENTIX_API` is `https://agentix.cloud` or unset): Run the registration flow (Steps 1–
The SaaS path explicitly selects agentix.cloud and directs the agent into its remote registration flow.
低
Hardcoded URL
AGENTIX_API_URL=http://localhost:3456
The localhost URL configures a self-hosted network service used by the skill. It confirms network capability even though the example remains on the host.
低
Hardcoded URL
- **SaaS (default):** `https://agentix.cloud` — multi-tenant, requires registration and API key. Fol
The line identifies agentix.cloud as the default multi-tenant service and requires remote registration and API authentication.
低
Hardcoded URL
- **Self-hosted:** Single team, no auth required. Set `AGENTIX_API_URL` to your instance URL (e.g.,
The line directs the skill to a user-selected self-hosted API. This is legitimate but confirms network access to a configurable service.
低
Hardcoded URL
These steps apply only to `https://agentix.cloud`. Self-hosted users skip this section.
The line scopes the following workflow to agentix.cloud, confirming intended use of that external service.
低
Hardcoded URL
POST https://agentix.cloud/register
The instruction sends a registration POST containing the user's name and email to agentix.cloud.
低
Hardcoded URL
GET https://agentix.cloud/register/<token>
The instruction polls agentix.cloud with a registration token to retrieve account credentials.
低
Hardcoded URL
POST https://agentix.cloud/teams
The instruction creates team data through an external agentix.cloud API request.
低
Hardcoded URL
PATCH https://agentix.cloud/teams/$TEAM_ID
The instruction modifies a team through agentix.cloud while using an authorization credential.
低
Hardcoded URL
{ "config": { "gitRepoUrl": "https://github.com/org/repo", "githubToken": "$GITHUB_TOKEN" } }
The configuration includes a GitHub repository URL, confirming integration with an external platform. Token exposure is covered by separate findings.
低
Hardcoded URL
GET https://agentix.cloud/docs/getting-started # human getting started guide
The skill directs a GET request to the public agentix.cloud getting-started documentation endpoint.
低
Hardcoded URL
GET https://agentix.cloud/docs/api-reference # full API reference
The skill directs a GET request to the public agentix.cloud API reference endpoint.
低
Hardcoded URL
GET https://agentix.cloud/docs # interactive API explorer (Swagger)
The skill directs a GET request to the public agentix.cloud interactive documentation endpoint.
低
Hardcoded URL
GET https://agentix.cloud/openapi.json # OpenAPI spec
The skill directs a GET request to the public agentix.cloud OpenAPI endpoint.
风险因素
⚙️ 外部命令 (50)
SKILL.md:14 SKILL.md:16 SKILL.md:17 SKILL.md:19-25 SKILL.md:25-27 SKILL.md:27-31 SKILL.md:31-37 SKILL.md:37 SKILL.md:38-40 SKILL.md:40-53 SKILL.md:53-62 SKILL.md:62-69 SKILL.md:69-73 SKILL.md:73-74 SKILL.md:74 SKILL.md:80-86 SKILL.md:86-91 SKILL.md:91-93 SKILL.md:93-95 SKILL.md:95-99 SKILL.md:99-101 SKILL.md:101-103 SKILL.md:103 SKILL.md:105 SKILL.md:109-114 SKILL.md:114-116 SKILL.md:116 SKILL.md:122-127 SKILL.md:127-139 SKILL.md:139-141 SKILL.md:141-143 SKILL.md:143 SKILL.md:150-159 SKILL.md:159-163 SKILL.md:163-167 SKILL.md:167-171 SKILL.md:171-176 SKILL.md:176-178 SKILL.md:178-186 SKILL.md:186-189 SKILL.md:189-192 SKILL.md:192-194 SKILL.md:194-197 SKILL.md:197-199 SKILL.md:199-203 SKILL.md:203-208 SKILL.md:208-210 SKILL.md:210-212 SKILL.md:212 SKILL.md:218
🌐 网络访问 (17)
📁 文件系统访问 (12)
审计者: codex 查看审计历史 →
分享与引用此报告
分享版本化评估报告、中性徽章、嵌入卡片和引用。Skillstore 只报告证据，不判断该 Skill 是否安全。
复制报告链接
https://skillstore.io/skills/agentix-cloud-agentix-ceo/audits/5?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_reportMarkdown 徽章
[](https://skillstore.io/skills/agentix-cloud-agentix-ceo?utm_source=security_passport_badge)HTML 徽章
<a href="https://skillstore.io/skills/agentix-cloud-agentix-ceo?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/agentix-cloud-agentix-ceo/security.svg" alt="Skillstore security assessment" loading="lazy"></a>嵌入卡片
<iframe src="https://skillstore.io/embed/skills/agentix-cloud-agentix-ceo.html" title="Skillstore Security Assessment" sandbox="allow-popups allow-popups-to-escape-sandbox" loading="lazy" referrerpolicy="no-referrer" width="420" height="180"></iframe>学术引用 (APA · BibTeX · CFF)
APA 引用
agentix-cloud. (2026). agentix-ceo security audit report (audit version 5) [Author version 1.2.0]. Skillstore. https://skillstore.io/skills/agentix-cloud-agentix-ceo/audits/5BibTeX 引用
@techreport{agentix-cloud-agentix-cloud-agentix-ceo-2026,
author = {agentix-cloud},
title = {agentix-ceo security audit report (audit version 5)},
institution = {Skillstore},
year = {2026},
number = {5},
url = {https://skillstore.io/skills/agentix-cloud-agentix-ceo/audits/5},
note = {Author version 1.2.0}
}CITATION.cff
cff-version: 1.2.0
message: "If you use this Skill, cite its author and this versioned security audit report."
title: "agentix-ceo security audit report (audit version 5)"
version: "1.2.0"
type: report
authors:
- name: "agentix-cloud"
date-released: "2026-07-23"
url: "https://skillstore.io/skills/agentix-cloud-agentix-ceo/audits/5"
identifiers:
- type: other
value: "skillstore:agentix-cloud-agentix-ceo:audit:5"
description: "Skillstore immutable audit report identifier"
Skillstore 评分
评分依据 证据置信度: 中 55
架构
100
可维护性
87
内容
67
社区
91
规范符合性
你能构建什么
协调工程工作
规划角色、创建任务，并启动 Agentix workers 以进行实现和审核工作。
运营自托管 Agentix 团队
将助手连接到本地 Agentix 实例，并管理角色、任务、workers 和事件。
跟踪多代理进展
检查活跃 Agentix 团队的任务状态、worker 活动和近期事件。
试试这些提示
设置 Agentix
设置我的 Agentix 团队。只询问你需要的信息，然后在保存凭据前向我展示涉及的每个外部服务。
审核团队状态
检查我的 Agentix 团队状态。总结正在运行的 workers、受阻任务、近期事件，以及需要我批准的操作。
创建聚焦的工作项
为这个目标创建 Agentix 任务。尽可能复用现有角色，并在创建任务前解释每个角色选择。
运行受监督交付
运行受监督的 Agentix 交付循环。监控 workers，在适当时恢复失败的工作，并在创建新角色或任务前询问。
最佳实践
- 在完全理解角色、任务和凭据处理之前，使用 supervised 模式。
- 在保存凭据或发送第三方 tokens 前，审核每个外部 endpoint。
- 为 Agentix worker 配置使用最小权限的 GitHub 和 API tokens。
避免
- 当有足够的 scoped token 可用时，不要共享权限过宽的 personal access tokens。
- 不要在没有明确运营策略的 repositories 或 teams 上启用 autopilot。
- 不要将该技能指向不受信任的自托管 Agentix endpoint。
常见问题
此技能是否需要 Agentix Cloud？
不需要。它支持 Agentix Cloud 和用户配置的自托管 Agentix endpoints。
它可以创建并运行 workers 吗？
可以。该技能记录了 Agentix workers 的任务运行和恢复 endpoints。
它会处理 secrets 吗？
会。在某些流程中，它会请求 Agentix、Anthropic 和 GitHub 凭据。用户应先审核 token scope。
什么是 supervised 模式？
Supervised 模式会监控并推进现有工作，但在规划新工作或生成 workers 前会先询问。
什么是 autopilot 模式？
Autopilot 模式允许助手监控、规划、创建任务、生成 workers、审核、合并并持续循环。
这对敏感 repositories 安全吗？
请谨慎使用。该技能会将 tokens 和工作指令发送到 Agentix 服务，因此敏感工作需要经过审查的信任边界。
开发者详情
许可证
MIT
作者版本
v1.2.0
Skillstore 修订
r2
引用
9e952417e76879bc9d853e1b8b2cd6d6d8d4a1c2
维护新鲜度
2026/7/24
使用量
11 次下载 · 0 次浏览
文件结构
📄 SKILL.md