Back to skill

Security audit

临时邮箱服务

Security checks for vulnerabilities and agentic risk

Overview

This temporary email skill has expected API-key and email-reading behavior, but its instructions include unrelated API routing and inaccurate read-only claims that need review before installation.

Review this skill carefully before installing. The temp-mail API access is broadly coherent, but the unrelated Douyin/Xiaohongshu fallback instructions should be removed or explained, the read-only claims should be corrected, and users should only fetch mailbox contents when they intend to expose verification codes or other email data to the agent and MaxHub service.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Findings (1)

T08 · Insecure Dependencies

Warning
Location
README.md:14
Finding

Unpinned Package Execution in Installation Instructions

Content
View full analysis

Vulnerability Details

File Locations:

  • README.md:14
  • README_CN.md:14

Vulnerability Type: Unpinned third-party package execution
Risk Level: Medium

Vulnerable Code

bash
npx clawhub install maxhub-temp-mail

Technical Analysis

The installation instructions invoke clawhub through npx without specifying a reviewed package version, integrity hash, lockfile, registry, or publisher-verification mechanism. Consequently, the package selected and executed when a user follows these instructions may differ from the version available when this Skill was audited.

This creates a supply-chain trust dependency on the package registry, the package publisher account, and the unresolved current package release. If any of those components are compromised, a malicious package version could be delivered through the documented installation command.

The issue is only triggered when a user manually executes the documented command; the reviewed Skill does not automatically run it.

Attack Path

  1. An attacker compromises the clawhub publisher account, package registry distribution path, or a transitive dependency used by the resolved package.
  2. The attacker publishes or serves a malicious package version.
  3. A user follows the installation instructions and executes npx clawhub install maxhub-temp-mail.
  4. npx resolves and downloads the unpinned package version.
  5. Malicious package, lifecycle, dependency, or CLI code executes with the privileges of the user running the command.

Impact Assessment

Successful exploitation could permit arbitrary code execution under the installing user's account. Depending on that account's permissions and host configuration, the malicious package could access user-readable files, environment variables, credentials, network resources, and project data or modify files writable by that user.

The instructions do not require administrative privileges, so privilege escalation is not inherent. The maxim ...[truncated 78 chars]

Remediation
View remediation

Remediation Suggestions

  1. Pin clawhub to a specific reviewed version:
bash
npx clawhub@<verified-version> install maxhub-temp-mail
  1. Document the expected package registry, canonical publisher, and package identity.
  2. Where supported, verify package signatures, provenance attestations, or integrity hashes before execution.
  3. Review and update the pinned version through a controlled dependency-update process rather than resolving the latest release at installation time.
  4. Apply the same corrected command to both README.md and README_CN.md.
  5. Consider using an installation mechanism backed by a lockfile or other reproducible dependency manifest.
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • System Prompt LeakageDirect Leakage, Indirect Extraction, Tool-Based Exfiltration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Findings (14)

Hidden Instructions

High
Category
Prompt Injection
Confidence
95% confidence
Finding

This duplicated hidden-instructions finding points to the same concealed path-selection guidance and unrelated API conventions. Such instructions can bypass clear interface boundaries and encourage the agent to follow opaque references instead of a narrowly reviewed temp-mail action set.

Content

Scanner excerpt · SKILL.md (reported line 69)May include surrounding context.

md
| 禁止行为 | 正确做法 |
|----------|----------|
| ❌ 自行拼接路径(如 `/api/v1/douyin/search/xxx`) | ✅ 使用 Action Table 或 `<!-- Full path: -->` 中的路径 |
| ❌ 猜测参数名(如把 `aweme_id` 写成 `video_id`) | ✅ 使用 Action Table 或 reference 文件中的参数名 |
| ❌ 假设 v1/v2/v3 参数兼容 | ✅ 降级时重新读取对应版本的参数文档 |
| ❌ 调用 `fetch_hot_search_list` 或 `app/v3/fetch_video_comments` | ✅ 使用替代端点(见废弃标注) |

Hidden Instructions

High
Category
Prompt Injection
Confidence
95% confidence
Finding

This duplicated hidden-instructions finding points to the same concealed path-selection guidance and unrelated API conventions. Such instructions can bypass clear interface boundaries and encourage the agent to follow opaque references instead of a narrowly reviewed temp-mail action set.

Content

Scanner excerpt · SKILL.md (reported line 69)May include surrounding context.

md
| 禁止行为 | 正确做法 |
|----------|----------|
| ❌ 自行拼接路径(如 `/api/v1/douyin/search/xxx`) | ✅ 使用 Action Table 或 `<!-- Full path: -->` 中的路径 |
| ❌ 猜测参数名(如把 `aweme_id` 写成 `video_id`) | ✅ 使用 Action Table 或 reference 文件中的参数名 |
| ❌ 假设 v1/v2/v3 参数兼容 | ✅ 降级时重新读取对应版本的参数文档 |
| ❌ 调用 `fetch_hot_search_list` 或 `app/v3/fetch_video_comments` | ✅ 使用替代端点(见废弃标注) |

Intent-Code Divergence

High
Category
Not specified by scanner
Confidence
96% confidence
Finding

The security declaration claims the skill is a read-only query tool, yet the skill explicitly supports temporary mailbox generation. Creating resources is not read-only, so this misrepresentation can mislead users, operators, and policy checks about what the skill actually does.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The skill is presented as a temp-mail assistant, but large sections route to unrelated Douyin/Xiaohongshu-style APIs and version-downgrade logic. This mismatch is dangerous because it can silently expand the skill’s operational scope, confuse reviewers, and enable unintended or unauthorized calls beyond the declared purpose.

Content

No source excerpt is available for this finding.

Direct Prompt Extraction

High
Category
System Prompt Leakage
Confidence
85% confidence
Finding

Skill contains instructions that could directly expose system prompts, internal rules, or hidden instructions to users or external parties.

Content

Scanner excerpt · SKILL.md (reported line 176)May include surrounding context.

md
| "analyze" / "分析一下" | Switch to analyze mode |
| "compare with X" / "和X对比" | Add X as second query |

## Output Guidelines

1. **Language consistency** — ALL output matches user's detected language.
2. **Markdown links** — All URLs in `[text](url)` format.

Rp1

Medium
Category
MCP Rug Pull
Confidence
89% confidence
Finding

The installation example uses npx clawhub install maxhub-temp-mail without pinning a specific version of the package being fetched and executed. This creates a supply-chain risk: users may execute a newly published or compromised package version, and because installation commands are commonly copied verbatim from README files, this can lead to unintended code execution on the user's system.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The example trigger phrases are extremely broad (邮箱, 邮件), overlapping with normal user conversation about email. In an agent environment, this can cause unintended invocation of the temp-mail skill during unrelated discussions, which is particularly risky because the skill handles disposable email generation and message retrieval that may expose message contents or create unexpected external side effects.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
91% confidence
Finding

The installation command uses npx clawhub without pinning a specific version, which can cause users to fetch and execute whatever package version is current at install time. This creates a supply-chain risk: a malicious or compromised upstream release could be executed directly on the user's system during installation.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The example trigger words 邮箱,邮件 are extremely broad everyday terms, so the skill may be invoked unintentionally during normal conversation about email. In this skill's context, accidental activation could expose mailbox metadata or cause unintended calls to the temporary mail service, especially if an API key is already configured.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
88% confidence
Finding

The skill sends user-driven data and mailbox queries to an external third-party service using an API key. External transmission is expected for API-backed skills, but here it is security-relevant because temp-mail usage may involve sensitive email addresses, message metadata, or contents being transmitted off-platform.

Content

Scanner excerpt · SKILL.md (reported line 53)May include surrounding context.

md
maxhub_auth_header="Authorization: Bearer ${MAXHUB_API_KEY}"

# GET example
curl -s "https://www.aconfig.cn/api/v1/temp-mail/{endpoint}?{params}" \
  -H "$maxhub_auth_header"

# POST example

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The trigger phrases are broad and overlap with common terms like 'get', 'mail', 'find', and 'show me', which increases the chance of accidental routing or invocation. In a skill that can create temp mailboxes and retrieve messages, unintended activation can expose sensitive workflows or cause unnecessary external requests.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

This documentation describes endpoints for generating temp mailboxes, listing emails, and retrieving email contents, but it does not warn that these operations expose potentially sensitive message data. In a temp-mail skill, lack of explicit disclosure and handling guidance can lead downstream agents or users to access, summarize, store, or relay mailbox contents without appropriate consent or caution.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
95% confidence
Finding

The instruction says ALL output must match the user's detected language, which imposes a language choice automatically rather than offering the user a choice. This is a natural-language policy concern because it fixes locale behavior without explicit opt-in or a documented exception.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
77% confidence
Finding

The file title and response schema explicitly include Chinese-language content such as 邮箱操作接口 and message_zh, but there is no statement about locale selection, opt-in, or a justified region-specific constraint. This can indicate a language/locale policy issue because the skill appears to impose or assume a specific locale without documenting user choice.

Content

No source excerpt is available for this finding.