技能 treg
📦
treg
v0.11.0 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
Access Live Data and Connected APIs with treg
Agents often lack credentials for live data and external APIs. treg searches priced endpoints and calls approved services through one authenticated interface.
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 "treg" from https://skillstore.io/skills/superdesigndev-treg.md and its manifest at https://skillstore.io/api/skills/superdesigndev-treg/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“treg”。 Find a provider for subreddit posts, but do not spend any balance.
预期结果:
I found matching endpoints and compared their required inputs, observed reliability, speed, and price. No endpoint was called.
正在使用“treg”。 Prepare a company enrichment request and use the least expensive reliable provider.
预期结果:
The selected provider accepts the company domain you have and has stronger observed reliability. The listed call price requires your confirmation.
正在使用“treg”。 Create a plan for a limited CI agent that can read Stripe balances.
预期结果:
Use a dedicated agent identity, allow only the Stripe tool, grant read-only access, set a daily cap, and review the audit log.
安全审计
严重v1 • 2026/8/14 打开版本化报告
The audit confirms remote pipe-to-shell installation, credential collection, hidden configuration access, and broad authenticated command execution. Many backtick and URL matches are documentation-only false positives, but the confirmed critical and high-risk workflows require strong user consent and verification.
1
已扫描文件
271
分析行数
34
复核项
0
已忽略误报
已确认安全问题 (6)
严重
Pipe to shell pattern
curl -fsSL https://treg.to/install.sh | sh # 1. the CLI (skip if `treg --version` already works)
The skill explicitly pipes an unpinned HTTPS response into sh, allowing remote code to execute without prior inspection.
严重
Pipe to shell pattern
curl -fsSL https://treg.to/install.sh | sh # installs the CLI + points it here
The repeated installation command pipes a mutable remote script directly into sh.
高
Environment file access
**Bulk (the fast path):** point treg at a directory — it detects provider keys in the `.env` AND
The bulk workflow scans a chosen directory for provider keys in .env and then registers selected credentials remotely.
高
Approval Weakening Instruction
The skill tells agents to request blanket approval for treg and avoid per-command confirmation, even though later commands upload secrets and change accounts.
The instruction explicitly labels additional confirmation as friction and asks for treg approval as a whole.
高
Third-Party Credential Collection and Custody
The workflows discover environment keys, read secret files or CLI keychains, and send credentials or OAuth material to treg for storage and injection.
The text directly describes .env detection, secret upload, keychain extraction, and hosted OAuth token capture.
高
Broad Credentialed Action Surface
The proxy relays arbitrary HTTP methods, headers, cookies, and bodies, while treg run executes vendor CLIs locally or remotely with injected credentials.
The documented interface supports state-changing HTTP requests and arbitrary vendor CLI arguments under organizational credentials.
能力复核项 (34)
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
Ruby/shell backtick execution
```bash
The fenced block begins an installation sequence that downloads a remote script, signs in, and changes agent configuration.
严重
Ruby/shell backtick execution
curl -fsSL https://treg.to/install.sh | sh # 1. the CLI (skip if `treg --version` already works)
This command downloads an unpinned remote script and immediately executes it, followed by login and MCP installation.
严重
Ruby/shell backtick execution
```bash
This fenced block installs remote code and performs authentication, including a token-based non-interactive login option.
严重
Hardcoded URL
curl -fsSL https://treg.to/install.sh | sh # 1. the CLI (skip if `treg --version` already works)
The URL supplies an installation script that is executed immediately without a pinned digest or signature verification.
严重
Hardcoded URL
curl -fsSL https://treg.to/install.sh | sh # installs the CLI + points it here
The external URL is the source of a remote installer that is piped directly into a shell.
高
Ruby/shell backtick execution
shell commands, request `treg` **once**, as a whole, rather than a prompt per subcommand. Halting
The skill asks for blanket approval of every treg subcommand, reducing oversight for later commands that spend funds, upload secrets, or change accounts.
高
Ruby/shell backtick execution
- "call it" → `call` with the endpoint id, or `<tool-name>/<path>` for one of the team's own tools
The call tool can invoke catalog endpoints or team tools, including state-changing upstream API operations with injected credentials.
高
Ruby/shell backtick execution
**Run a registered CLI tool** — the command-line complement to `treg call` (which proxies HTTP
The skill directs execution of registered vendor CLI tools with organizational credentials, creating a broad command and account-action surface.
高
Ruby/shell backtick execution
APIs). `treg run <tool> -- <cli args>` runs a vendor CLI (stripe, gh, vercel, gcloud...) with the ke
treg run forwards arbitrary CLI arguments to vendor tools and injects credentials, enabling authenticated operations chosen at runtime.
高
Ruby/shell backtick execution
injected, so you never hold or log into it. Two tiers: `--local` (default, runs on this machine) ·
The documented local tier executes vendor commands on the user's machine, while the server tier executes them remotely with credentials.
高
Ruby/shell backtick execution
```
The fenced examples execute Stripe locally and another credentialed CLI on the registry server.
高
Ruby/shell backtick execution
treg upload # both sides of the cwd; `treg upload env|skills --dir <d>` to res
The upload command scans the working directory and can register selected environment secrets and skills with the external treg service.
高
Ruby/shell backtick execution
```bash
The fenced sequence creates local files, initializes a skill, and registers its recipe, secret, and tool with the remote service.
高
Ruby/shell backtick execution
Call it: `treg call posthog api/projects/@current` (upstream: https://us.posthog.com)
This range includes commands that initialize and remotely register a skill bundle containing a recipe, secret, and API tool.
高
Ruby/shell backtick execution
```bash
The fenced commands read an API key or secret file, upload the secret, and register authenticated upstream tools.
高
Ruby/shell backtick execution
```bash
The fenced sequence scans a hidden Claude skill directory, captures secret files, and pushes the resulting credential bundle remotely.
高
Ruby/shell backtick execution
- **Manual:** do your own OAuth locally, then `treg secret add gsc --file token.json --kind oauth`.
The command uploads a locally obtained OAuth token file to treg for storage and later injection.
高
Ruby/shell backtick execution
- **Hosted connect:** `treg oauth connect gsc --client-secret client_secret.json --scopes <scope>`
The hosted OAuth command supplies a client-secret file to treg and delegates token capture to its callback service.
高
Ruby/shell backtick execution
```bash
The fenced administrative commands can remove tools, secrets, and skills or actively refresh and probe credentials.
高
Ruby/shell backtick execution
```bash
The fenced commands create organizations, invite members, and switch account context, causing consequential remote state changes.
高
Ruby/shell backtick execution
```bash
The fenced commands mint, scope, rotate, and revoke agent authentication tokens, which are security-sensitive account operations.
高
Hardcoded URL
One-time setup: add `https://treg.to/oauth/callback` to your OAuth app's redirect URIs.
The OAuth redirect sends authorization results to treg, allowing the service to capture and retain delegated tokens.
高
Hidden file in home directory
this same skill into `~/.claude/skills/treg/`. It is harmless, but it duplicates what this plugin
The installer writes a skill copy into the hidden ~/.claude configuration tree, modifying agent behavior outside the current workspace.
高
Hidden file in home directory
treg skill scaffold ~/.claude/skills/google-ads --out gads.json
The command scans a hidden Claude skill directory and captures its recipe and secret files into a bundle.
高
Hidden file access
this same skill into `~/.claude/skills/treg/`. It is harmless, but it duplicates what this plugin
The setup modifies the hidden ~/.claude skills directory and explicitly leaves a duplicate skill installation there.
高
Hidden file access
treg secret add posthog-key --value "$POSTHOG_API_KEY" # or --file ./.secret/token.json
The command can read a token from ./.secret/token.json and send it to treg for storage.
高
Hidden file access
treg skill scaffold ~/.claude/skills/google-ads --out gads.json
The scaffold command traverses ~/.claude/skills/google-ads and captures every .secret file described by the following lines.
高
Generic API/secret keys
treg secret add posthog-key --value "$POSTHOG_API_KEY" # or --file ./.secret/token.json
The command reads POSTHOG_API_KEY from the environment and uploads its value to the external treg service.
中
Ruby/shell backtick execution
email — see them with `treg invites`, accept with `treg accept` (or `treg org join <code>`). Switch
The documented commands accept organization invitations, join organizations, and switch the active organization, which changes account state.
中
Ruby/shell backtick execution
teams: `treg org switch <slug>`.
The command changes the active organization used by all later operations, so accidental execution can affect the wrong account.
中
Ruby/shell backtick execution
```bash
The fenced commands invoke external catalog APIs and can spend the team's prepaid balance, although the skill also requires price disclosure.
中
Ruby/shell backtick execution
The invitee signs in with the invited email and runs `treg accept` — no code handling needed
The accept command changes organization membership for the signed-in account.
中
Ruby/shell backtick execution
(the code path still works: `treg org join <code>`). A brand-new invitee also gets their own
The join command uses an invitation code to change organization membership and account context.
中
Hardcoded URL
through treg: GET https://treg.to/call/https://api.intercom.io/conversations?per_page=5
The example routes an upstream API request through treg, exposing request metadata and content to an additional external service.
风险因素
⚙️ 外部命令 (50)
SKILL.md:9 SKILL.md:12-13 SKILL.md:13-16 SKILL.md:16-20 SKILL.md:20 SKILL.md:28-49 SKILL.md:49 SKILL.md:50-58 SKILL.md:58-59 SKILL.md:59 SKILL.md:63-67 SKILL.md:67 SKILL.md:68 SKILL.md:71-76 SKILL.md:76-78 SKILL.md:78 SKILL.md:79-83 SKILL.md:83-84 SKILL.md:84 SKILL.md:87 SKILL.md:88 SKILL.md:89-92 SKILL.md:92 SKILL.md:95 SKILL.md:106-112 SKILL.md:112-114 SKILL.md:114-116 SKILL.md:116-117 SKILL.md:117 SKILL.md:122-123 SKILL.md:123 SKILL.md:124 SKILL.md:125-127 SKILL.md:127-128 SKILL.md:128 SKILL.md:130 SKILL.md:131-143 SKILL.md:143-144 SKILL.md:144-145 SKILL.md:145-155 SKILL.md:155-159 SKILL.md:159-162 SKILL.md:162-164 SKILL.md:164 SKILL.md:165-167 SKILL.md:167-168 SKILL.md:168-169 SKILL.md:169-170 SKILL.md:170 SKILL.md:171-174
🌐 网络访问 (11)
📁 文件系统访问 (5)
🔑 环境变量 (2)
检测到的模式
Pipe to shell pattern×2
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/superdesigndev-treg/audits/1?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_reportMarkdown 徽章
[](https://skillstore.io/skills/superdesigndev-treg?utm_source=security_passport_badge)HTML 徽章
<a href="https://skillstore.io/skills/superdesigndev-treg?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/superdesigndev-treg/security.svg" alt="Skillstore security assessment" loading="lazy"></a>嵌入卡片
<iframe src="https://skillstore.io/embed/skills/superdesigndev-treg.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
superdesigndev. (2026). treg security audit report (audit version 1) [Author version 0.11.0]. Skillstore. https://skillstore.io/skills/superdesigndev-treg/audits/1BibTeX 引用
@techreport{superdesigndev-superdesigndev-treg-2026,
author = {superdesigndev},
title = {treg security audit report (audit version 1)},
institution = {Skillstore},
year = {2026},
number = {1},
url = {https://skillstore.io/skills/superdesigndev-treg/audits/1},
note = {Author version 0.11.0}
}CITATION.cff
cff-version: 1.2.0
message: "If you use this Skill, cite its author and this versioned security audit report."
title: "treg security audit report (audit version 1)"
version: "0.11.0"
type: report
authors:
- name: "superdesigndev"
date-released: "2026-08-14"
url: "https://skillstore.io/skills/superdesigndev-treg/audits/1"
identifiers:
- type: other
value: "skillstore:superdesigndev-treg:audit:1"
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
65
社区
91
规范符合性
你能构建什么
Research live market data
Find and compare endpoints for rankings, backlinks, trends, advertising, or company enrichment before making approved calls.
Connect internal API accounts
Register an existing service credential and let authorized teammates call the API without receiving the raw secret.
Automate scoped agent tasks
Create a limited agent identity, allow selected tools, set daily caps, and review its call history.
试试这些提示
Find a data endpoint
Search treg for an endpoint that provides [data type]. Show required inputs, price, and reliability. Do not call it yet.
Compare providers
Compare treg providers for [task] using my available inputs. Rank them by compatibility, reliability, price, and recent success.
Run an approved catalog request
Prepare a treg call for [endpoint] with [inputs]. State the exact price and request confirmation before spending any balance.
Design a restricted team integration
Plan a treg integration for [service]. Use minimal scopes, read-only access, explicit tool allowlists, daily caps, and item-level secret consent.
最佳实践
- Review endpoint inputs, observed reliability, price, and recent success before calling.
- Confirm every paid or state-changing action with the user immediately before execution.
- Use dedicated identities, minimal scopes, tool allowlists, daily caps, and audit logs.
避免
- Do not pipe a remote installer into a shell without verification.
- Do not grant blanket approval for every treg command.
- Do not scan or upload credentials without an exact preview and item-level consent.
常见问题
What does treg provide?
It provides a searchable API catalog, MCP tools, a CLI, and access to team-registered tools with credential injection.
Does every call cost money?
No. Catalog calls served with treg credentials can spend prepaid balance. Calls using a team's own registered credentials are described as unmetered.
Can treg access existing team APIs?
Yes. Administrators can register API bases and credentials, then authorized agents can proxy requests through treg.
Does treg handle secrets?
Yes. It can store environment values, token files, OAuth material, and CLI authentication for later server-side injection.
Can treg change connected accounts?
Yes. Proxied HTTP methods and vendor CLI commands can perform state-changing actions when the connected credential permits them.
What should users verify before installation?
Verify the installer source and integrity, review requested permissions, and understand credential storage, retention, and deletion policies.
开发者详情
许可证
MIT
作者版本
v0.11.0
Skillstore 修订
r1
引用
8d471fa6bb01c43c43d332bf9105df157c067391
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