Back to skill

Security audit

灵造

Security checks for vulnerabilities and agentic risk

Overview

This is a disclosed creator-research and content workflow skill that uses a Lingzao API key for online lookups and image generation, with install/update cautions but no evidence of hidden or malicious behavior.

Install only if you are comfortable giving this skill a Lingzao API key and letting it send requested public-content queries, prompts, and optional reference images to the configured Lingzao service. Review the base URL and update source before setup, and avoid running floating update commands in sensitive environments unless you trust the publisher and release channel.

Vulnerability Patterns
  • 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
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • System Prompt LeakageDirect Leakage, Indirect Extraction, Tool-Based Exfiltration
  • Memory PoisoningPersistent Context Injection, Context Window Stuffing, Memory Manipulation
  • Rogue AgentSelf-Modification, Session Persistence
Findings (97)

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

声明描述的是一个用于内容创作研究和自媒体运营的功能型 Skill,核心能力应围绕选题、写作、数据查询、账号诊断等展开。而给出的代码并未实现任何创作者研究、自媒体运营、内容生成、平台查询或 API Key 相关能力;它只是一个开发/测试辅助脚本,用于验证 playbook 路由配置是否正确。这属于 materially different primary purpose,且实际能力与声明功能基本无关,因此应判定为描述与行为不匹配。

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The declared description describes a feature-rich creator research and content-operations skill with multiple analysis, querying, writing, and generation capabilities. The actual supplied code chunk only reads a local VERSION file and prints the version string. This is a materially different primary purpose from the declared functionality and does not implement or evidence the described capabilities. While this may be a harmless support script within a larger project, based on the supplied code chunk alone, the behavior does not match the declared purpose.

Content

No source excerpt is available for this finding.

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 360)May include surrounding context.

md
Resolve this `SKILL.md` directory as `<skill_root>`, then run setup once:

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 401)May include surrounding context.

md
Resolve this `SKILL.md` directory as `<skill_root>`, then run setup once:

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 401)May include surrounding context.

md
s not exist, find the directory that contains `lingzao`'s `SKILL.md`, then run `scripts/setup.sh --skip-doctor` from that directory.

External Script Fetching

High
Category
Supply Chain
Confidence
94% confidence
Finding

The skill tells the agent to curl -L a URL returned by external data into a local file without any origin validation, content-type checks, size limits, or integrity enforcement. This creates a direct untrusted download path that could be abused for SSRF-like access, malicious content retrieval, oversized downloads, or feeding hostile data into later analysis steps.

Content

Scanner excerpt · SKILL.md (reported line 600)May include surrounding context.

Do not search only inside items[]. If data.artifacts.subtitle_markdown.status == "ready" and url exists, download it before deep script or subtitle analysis:

bash
curl -L "$subtitle_markdown_url" -o /tmp/lingzao-profile-subtitles.md

Use the downloaded Markdown file for complete subtitle/copy analysis. Use --format json when the user needs the structured fields. JSON includes data.artifacts.subtitle_markdown.url for the complete Markdown file when available, and inline items[].text.subtitle.content/plain_text are preview-sized to keep the response readable. If the artifact is unavailable, use the inline subtitle fields. For Douyin, expect data.artifacts.subtitle_markdown.status == "unsupported" and use the returned profile insights plus selected-video extraction instead.

Memory Manipulation

High
Category
Memory Poisoning
Confidence
80% confidence
Finding

Skill manipulates agent memory, state, or stored context. Memory corruption can alter personality, override safety rules, or cause unpredictable behavior.

Content

Scanner excerpt · playbooks/image-generation-agent-integration-guide.md (reported line 90)May include surrounding context.

md
- platform/format or aspect ratio
- visual direction: reference image, color palette, style group, or real
  material
- exact on-image text or a clear statement that Lingzao should choose the text

If the minimum brief is not met, return `next_action: send_reference_or_color`
or a similar friendly state instead of starting generation.

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 · playbooks/image-generation-execution-workflow.md (reported line 19)May include surrounding context.

md
## Product Principle

Do not expose prompt-only work as the main product experience.

User-facing output should be:

Anti-Refusal Statement

High
Category
Anti-Refusal
Confidence
85% confidence
Finding

Skill instructs the agent to omit warnings, disclaimers, or ethical commentary. Stripping safety caveats hides risk from the user and is a common jailbreak preamble.

Content

Scanner excerpt · playbooks/draft-rewrite-and-benchmark-workflow.md (reported line 274)May include surrounding context.

md
## Core Rule

Do not judge a missing draft. First confirm whether the user has finished the
content.

If the user has not sent the content yet, use short wording:

Anti-Refusal Statement

High
Category
Anti-Refusal
Confidence
85% confidence
Finding

Skill instructs the agent to omit warnings, disclaimers, or ethical commentary. Stripping safety caveats hides risk from the user and is a common jailbreak preamble.

Content

Scanner excerpt · playbooks/pre-publish-readiness-check.md (reported line 13)May include surrounding context.

md
## Core Rule

Do not judge a missing draft. First confirm whether the user has finished the
content.

If the user has not sent the content yet, use short wording:

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 · scripts/lingzao_client.py (reported line 821)May include surrounding context.

python
if not prompt.strip():
        raise LingzaoError("generate-image prompt cannot be empty.")
    return prompt


def check_skill_version(skill_base_url: str, timeout: int = DEFAULT_TIMEOUT) -> dict:

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
95% confidence
Finding

The skill exposes broad capabilities including shell, network, file read/write, and environment access, but does not declare an explicit tool scope or permission boundary. In an agent runtime, that creates unnecessary ambient authority: a prompt-routed workflow can invoke powerful operations without a clear allowlist, increasing the blast radius of prompt injection, accidental misuse, or unsafe playbook content.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
60% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · SKILL.md (reported line 213)May include surrounding context.

md
If the draft contains off-platform diversion, WeChat/private-contact guidance,
incentivized comment interaction, exaggerated guarantees, or sensitive
unsupported claims, do not leave those lines in the publishable version. Show a
short risk note and rewrite them into a safer Xiaohongshu version. Never
promise platform approval; say the rewrite lowers risk.

For commercial or product-related Xiaohongshu outputs, keep the order:

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The file prescribes exact Chinese text to use when a user has installed the skill but not configured an API key. There is no accompanying instruction to detect or honor the user's preferred language, so this creates a language-policy issue for non-Chinese users.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
90% confidence
Finding

The setup flow instructs use of npx skills without pinning an exact package version. That makes installation dependent on whatever version is current at execution time, enabling supply-chain drift or malicious upstream updates to change behavior unexpectedly in a privileged install/update path.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
90% confidence
Finding

The update instructions again rely on unpinned npx skills, so future updates may silently introduce different code than the reviewed version. In a skill that can run shell commands and write files, this increases supply-chain risk during maintenance as well as initial install.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
86% confidence
Finding

The skill instructs the agent to materialize user-supplied images into per-run temp directories and optionally create resized copies on disk. Even though it says not to store them in the repo, it still normalizes and persists user-provided content locally, which can expose sensitive images to other local users/processes if temp storage is predictable, long-lived, or insufficiently permissioned.

Content

Scanner excerpt · SKILL.md (reported line 794)May include surrounding context.

md
- Do not proactively convert image formats. If the input image is already png,
  jpeg, or webp and its file size is reasonable, pass it as-is. Do not convert
  png to webp or jpeg just because an example path uses a different extension.
- Only when a reference image is larger than 2 MB, create a smaller copy in the
  temp directory and pass that copy with `--image`. Keep the file extension and
  actual image bytes consistent. If resizing or compression fails, use the
  original supported image file instead of trying another format.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

The default prompt repeatedly instructs the skill to produce Xiaohongshu-facing deliverables and applies platform-specific Chinese content rules, while the skill branding and description are also Chinese-centric. There is no visible opt-in or user choice for language/locale, which may violate a policy requiring language or locale choice unless clearly justified and documented.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The manifest enables implicit invocation for a skill that carries a very broad default prompt and can trigger external lookups or generation flows. Without trigger constraints or exclusion conditions, the host agent may invoke this skill in contexts the user did not intend, causing over-collection of data, unintended outbound API use, or policy-shaping of unrelated tasks.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
84% confidence
Finding

The natural-language release note says the package 'keeps the Chinese WorkBuddy/SkillHub first screen and marketplace copy'. That indicates a language choice being enforced at the skill/package level, and this file does not present it as optional, user-selectable, or region-justified.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The skill provides required user-facing prompt wording only in Chinese, and the examples/instructions consistently direct responses in Chinese. There is no indication that the user can choose another language or that the locale restriction is justified as region-specific.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The activation rule is broad enough that it can trigger on vague statements like wanting to make money or not knowing what to do, causing the agent to steer users into this workflow even when they did not explicitly ask for account-direction analysis. In a multi-skill agent, that can lead to misrouting, unnecessary profiling questions, and over-collection of personal context before user intent is clear.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The keyword tree explicitly frames beginner users as 'especially women' and then uses women-centered assumptions as a default guidance path. This can cause biased personalization, inaccurate advice for users who did not opt into gendered framing, and inappropriate inference about sensitive attributes from sparse input.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

The 'Female Growth / 30+ / 35+' section presents gender-specific recommendations as a default starter map, which can steer users into demographic assumptions without consent. In conversational systems, this increases the risk of stereotyping and can skew downstream recommendations, searches, and monetization advice based on inferred gender rather than stated needs.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

Line L015 provides a fixed user-facing response entirely in Chinese as the prescribed answer. Similar mandatory Chinese response templates recur throughout the file, and the playbook does not indicate that the user can choose another language or that the locale restriction is optional or region-specific.

Content

No source excerpt is available for this finding.