技能 valet
📦
valet
内容修订 r1 严重 ⚙️ 外部命令🌐 网络访问📁 文件系统访问🔑 环境变量
Manage Valet Agents and Integrations
Building and operating Valet agents requires coordinated CLI, credential, connector, channel, and deployment steps. This skill guides those workflows with explicit confirmation, validation, and testing.
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 "valet" from https://skillstore.io/skills/valetdotdev-valet.md and its manifest at https://skillstore.io/api/skills/valetdotdev-valet/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“valet”。 Create a GitHub pull request assistant for the engineering organization.
预期结果:
A confirmed plan covering the agent project, GitHub connector, webhook channel, required token scopes, deployment, and pull request test.
正在使用“valet”。 My agent needs a different API credential from other organization agents.
预期结果:
An agent-scoped secret override plan that preserves the shared connector and limits credential exposure.
正在使用“valet”。 Capture this working session as a reusable agent.
预期结果:
- A session summary with detected tools, corrections, and workflow phases.
- A confirmation step before private session details become agent files.
- A reusable project with redacted placeholders and setup documentation.
安全审计
严重v1 • 2026/8/15 打开版本化报告
Most static matches are Markdown formatting, placeholders, or benign documentation examples rather than executable backticks or exposed credentials. Confirmed risks include unverified remote installer execution, broad organization-scoped credential sharing, private Claude session access, and predictable temporary log files. Static review was capped at 400/477 representative findings; omitted static matches are unconfirmed, so automatic publishing stays disabled until manual review.
1
已扫描文件
1,704
分析行数
13
复核项
0
已忽略误报
已确认安全问题 (2)
严重
Pipe to shell pattern
curl -fsSL https://valet.dev/install.sh | sh
The instruction downloads a remote script and executes it immediately through sh. A compromised host, DNS path, or installer can produce arbitrary code execution without integrity verification.
高
Private Session Content Reused in Deployable Artifacts
The capture workflow reads Claude session prompts and tool history, then uses that material to generate agent files. It does not require a dedicated sensitive-data review before deployment.
The instructions explicitly read hidden session records, extract user prompts, and reuse session content when generating an agent. Placeholder replacement reduces but does not eliminate disclosure risk.
能力复核项 (13)
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。
严重
Hardcoded URL
curl -fsSL https://valet.dev/install.sh | sh
This URL supplies a remote installer that the documented command executes immediately. Trusting mutable network content without a pinned digest or signature creates a supply-chain execution risk.
严重
Hardcoded URL
irm https://valet.dev/install.ps1 | iex
This URL supplies a remote installer that the documented command executes immediately. Trusting mutable network content without a pinned digest or signature creates a supply-chain execution risk.
高
Git platform tokens
3. **Org-scoped by default**: Always create connectors, channels, and env vars at the org level (`--
The guidance defaults credentials to organization scope, making one token usable through shared resources by multiple agents. This increases credential blast radius beyond least privilege.
高
Git platform tokens
**Default to `--org` when setting env vars.** Org-scoped env vars are available to every org-scoped
The guidance defaults credentials to organization scope, making one token usable through shared resources by multiple agents. This increases credential blast radius beyond least privilege.
高
Hidden file in home directory
`~/.claude/projects/-<cwd-with-slashes-replaced-by-dashes>/`
The workflow intentionally accesses hidden Claude project records containing session prompts and tool history. These private transcripts can contain sensitive user or project data.
高
Hidden file in home directory
Example: `/Users/me/Developer/my-project` → `~/.claude/projects/-Users-me-Developer-my-project/`
The workflow intentionally accesses hidden Claude project records containing session prompts and tool history. These private transcripts can contain sensitive user or project data.
高
Hidden file in home directory
ls -t ~/.claude/projects/-<path>/*.jsonl | head -1
The workflow intentionally accesses hidden Claude project records containing session prompts and tool history. These private transcripts can contain sensitive user or project data.
高
Hidden file in home directory
Also check `~/.claude/projects/<project-path>/sessions-index.json` for `summary` and `firstPrompt` f
The workflow intentionally accesses hidden Claude project records containing session prompts and tool history. These private transcripts can contain sensitive user or project data.
中
Hidden file access
`~/.claude/projects/-<cwd-with-slashes-replaced-by-dashes>/`
The workflow intentionally accesses hidden Claude project records containing session prompts and tool history. These private transcripts can contain sensitive user or project data.
中
Hidden file access
Example: `/Users/me/Developer/my-project` → `~/.claude/projects/-Users-me-Developer-my-project/`
The workflow intentionally accesses hidden Claude project records containing session prompts and tool history. These private transcripts can contain sensitive user or project data.
中
Hidden file access
ls -t ~/.claude/projects/-<path>/*.jsonl | head -1
The workflow intentionally accesses hidden Claude project records containing session prompts and tool history. These private transcripts can contain sensitive user or project data.
中
Hidden file access
Also check `~/.claude/projects/<project-path>/sessions-index.json` for `summary` and `firstPrompt` f
The workflow intentionally accesses hidden Claude project records containing session prompts and tool history. These private transcripts can contain sensitive user or project data.
中
Temp directory access
valet logs > /tmp/valet-test-<agent-name>.log 2>&1
The workflow writes agent logs to a predictable shared /tmp path without restrictive permissions or cleanup. This can expose logs or permit symlink-based file replacement on multi-user systems.
风险因素
⚙️ 外部命令 (50)
SKILL.md:7 SKILL.md:16 SKILL.md:18 SKILL.md:29-31 SKILL.md:31-35 SKILL.md:35-37 SKILL.md:37-39 SKILL.md:39-40 SKILL.md:40-47 SKILL.md:47-49 SKILL.md:49-60 SKILL.md:60-61 SKILL.md:61-68 SKILL.md:68-70 SKILL.md:70-72 SKILL.md:72-80 SKILL.md:80-86 SKILL.md:86-88 SKILL.md:88 SKILL.md:90-99 SKILL.md:99-101 SKILL.md:101-103 SKILL.md:103 SKILL.md:104 SKILL.md:106-109 SKILL.md:109 SKILL.md:113-114 SKILL.md:114-115 SKILL.md:115 SKILL.md:116 SKILL.md:117 SKILL.md:118 SKILL.md:120-124 SKILL.md:124-128 SKILL.md:128-129 SKILL.md:129 SKILL.md:130 SKILL.md:131 SKILL.md:133 SKILL.md:139 SKILL.md:140 SKILL.md:141 SKILL.md:142-144 SKILL.md:144-151 SKILL.md:151-153 SKILL.md:153-156 SKILL.md:156-158 SKILL.md:158-160 SKILL.md:160-161 SKILL.md:161-162
🌐 网络访问 (10)
📁 文件系统访问 (13)
检测到的模式
Pipe to shell pattern
审计者: codex
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/valetdotdev-valet/audits/1?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_reportMarkdown 徽章
[](https://skillstore.io/skills/valetdotdev-valet?utm_source=security_passport_badge)HTML 徽章
<a href="https://skillstore.io/skills/valetdotdev-valet?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/valetdotdev-valet/security.svg" alt="Skillstore security assessment" loading="lazy"></a>嵌入卡片
<iframe src="https://skillstore.io/embed/skills/valetdotdev-valet.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
valetdotdev. (2026). valet security audit report (audit version 1) [Author version unspecified]. Skillstore. https://skillstore.io/skills/valetdotdev-valet/audits/1BibTeX 引用
@techreport{valetdotdev-valetdotdev-valet-2026,
author = {valetdotdev},
title = {valet security audit report (audit version 1)},
institution = {Skillstore},
year = {2026},
number = {1},
url = {https://skillstore.io/skills/valetdotdev-valet/audits/1},
note = {Author version unspecified}
}CITATION.cff
cff-version: 1.2.0
message: "If you use this Skill, cite its author and this versioned security audit report."
title: "valet security audit report (audit version 1)"
version: "unspecified"
type: report
authors:
- name: "valetdotdev"
date-released: "2026-08-15"
url: "https://skillstore.io/skills/valetdotdev-valet/audits/1"
identifiers:
- type: other
value: "skillstore:valetdotdev-valet:audit:1"
description: "Skillstore immutable audit report identifier"
Skillstore 评分
评分依据 证据置信度: 低 55
架构
85
可维护性
87
内容
65
社区
70
规范符合性
你能构建什么
Deploy a Team Agent
Create an agent, attach shared connectors and channels, configure secrets, deploy it, and verify the complete workflow.
Build an Event Automation
Design a webhook, Slack, cron, or heartbeat agent with scoped channel instructions and tested connector access.
Capture a Repeatable Workflow
Transform a reviewed working session into reusable agent instructions, setup documentation, and deployment metadata.
试试这些提示
Inspect My Valet Setup
Check whether Valet is installed and authenticated. Then list my organizations and agents without making changes.
Create a Basic Agent
Create a Valet agent named <agent-name> in <organization>. Present the file and deployment plan before making changes.
Configure an Integration
Add a <service> connector and <channel-type> channel to <agent-name>. Prefer catalog resources and explain required credentials without requesting their values.
Design and Validate an Automation
Design an agent for <workflow>. Include triggers, scope boundaries, failure handling, least-privilege credentials, deployment steps, connector tests, and end-to-end validation.
最佳实践
- Review every command and target organization before execution.
- Use least-privilege secrets and test each protected connector before deployment.
- Validate manifests and complete an end-to-end channel test before declaring success.
避免
- Do not place real credentials in prompts, project files, examples, or command history.
- Do not run installation, destructive operations, or production test events without explicit approval.
- Do not share organization-wide credentials when an agent-scoped credential is sufficient.
常见问题
What is required before using this skill?
You need the Valet CLI. Account operations also require an authenticated Valet account.
Can the skill install the Valet CLI?
Yes, but it must explain the installer and receive explicit permission before running it.
Will the skill ask me to paste API keys?
No. It directs you to set secrets in your terminal and waits for confirmation.
Can it configure connectors and channels?
Yes. It checks Valet catalogs first, reuses existing resources, and creates custom resources only when needed.
Can it build an agent from a previous session?
Yes. Review and redact the extracted session summary before any generated project is deployed.
Does it confirm destructive operations?
Yes. Agent, connector, channel, organization, and session removal require explicit user confirmation.
开发者详情
作者
valetdotdev许可证
MIT
Skillstore 修订
r1
版本提示
作者未声明版本。
引用
60855416b9ec7b18f5e06fa97cdadcfda015db7a
维护新鲜度
2026/8/15
使用量
0 次下载 · 0 次浏览
文件结构
📄 SKILL.md