Security audit
Security checks for vulnerabilities and agentic risk
This is a disclosed skill-building assistant that can create or edit skill files with user confirmation, with no evidence of hidden execution, persistence, or data exfiltration.
Install this only if you want an assistant that helps create or modify skill files. Review generated SKILL.md and reference files before installing or publishing them, and avoid giving it secrets, credentials, private customer data, or unrelated sensitive material because it may summarize provided context into reusable skill artifacts.
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
# 本地凭证
config.local.json
config.local.*.json
.env
.env.local
# 临时脚本
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
# 本地凭证
config.local.json
config.local.*.json
.env
.env.local
# 临时脚本
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
config.local.json
config.local.*.json
.env
.env.local
# 临时脚本
_*.py
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
config.local.json
config.local.*.json
.env
.env.local
# 临时脚本
_*.py
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.
When user's answers contradict across rounds:
1. **Quote both answers verbatim**
2. **Present the contradiction without judgment**
3. **Let the user choose**
**Example**:
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.
When user's answers contradict across rounds:
1. **Quote both answers verbatim**
2. **Present the contradiction without judgment**
3. **Let the user choose**
**Example**:
The guidance requires README warnings to be presented in both Chinese and English, and examples prescribe Chinese section titles like "用户须知"/"⚠️ 注意". This imposes a specific language/locale policy on all skill authors without an explicit opt-in or a documented region-specific justification, which matches the language/locale policy violation category.
This markdown file contains natural-language guidance that assumes Chinese language usage in headings, workflow text, and trigger examples while the surrounding document is otherwise English. Under the policy rule, forcing or assuming a specific language/locale without explicit user opt-in is a natural-language policy concern.
This reference instructs the skill to ask users questions in Chinese, including quoted prompts such as "你想让这个 Skill 帮你做什么？" and confirmation text, but it does not state that the skill is Chinese-only or provide any user opt-in for language selection. That creates a natural-language policy concern because the interaction language is effectively forced by default.
The document mixes English framing with prescriptive Chinese example prompts and presents them as the default interviewing method rather than an optional locale. This can violate language-choice policy because it implicitly constrains interactions to Chinese without stating that the user may choose another language.
This markdown file contains instructions, examples, and output formatting entirely in Chinese, including fixed output labels and phrasing, with no indication that users may choose another language. That creates a natural-language policy concern because it effectively forces a specific language without user opt-in.
The document embeds core operational instructions, trigger examples, and diagnosis flows primarily in Chinese, including user-facing prompts and routing criteria. There is no indication that users may choose another language or that the skill is intentionally limited to a Chinese-speaking context, which creates a language/locale policy concern.
The route detection rules rely on short generic phrases such as "刚才/刚刚/存成/变成" and "不触发/不工作/跑偏/太啰嗦/改进" to infer entry routes. These phrases are common in ordinary conversation and the file does not provide enough exclusion conditions or negative examples to prevent unintended routing.
The document title and all instructional content are presented in Chinese, and line L065 further requires bilingual warnings, implying language expectations for skill authors. Under the policy rule, a skill artifact should not force a specific language or locale unless it offers user choice or clearly documents a justified regional constraint.
This markdown file contains substantial natural-language instructions in Chinese alongside English, including core workflow headings and terminology such as '腾讯9维度自评' and '任务/输出格式/规则/示例'. Under the policy, forcing a specific language or locale without user opt-in should be flagged unless the locale constraint is justified, and this file does not offer a language choice or explain why Chinese is required.
The document consistently instructs the skill to conduct interview prompts and adaptations in Chinese, including fixed Chinese question text and user-language heuristics, without offering a language choice. This creates a natural-language policy concern because it effectively forces a specific language/locale on users rather than making it optional.
The reference repeatedly prescribes interview questions in Chinese, beginning with the BAD/GOOD examples here and continuing throughout the file. This creates a language/locale constraint without offering user opt-in or explaining that the skill is intended only for a Chinese-language context.
The JSON example hard-codes Chinese question text, headers, labels, and descriptions as the prescribed interaction format. Because the document presents this as normative format guidance rather than a locale-specific example, it implies a fixed language policy that may conflict with users' preferences.
The file presents the skill entirely in Chinese and specifies fixed Chinese output labels and formatting, which implies a language requirement for use and output. Under the policy rule, forcing a specific language without user opt-in is a natural-language policy violation unless the locale restriction is explicitly justified.
The R2 dialog-extraction flow explicitly tells the agent to scan all conversation context, which can pull in prior messages containing secrets, unrelated personal data, or sensitive business information without any user-facing notice or minimization step. In a skill that turns conversations into reusable artifacts, broad context review increases the chance of over-collection and accidental propagation of sensitive data into the generated skill.
The R3 material-analysis route directs the agent to read supplied documents thoroughly but provides no safeguards for handling confidential documents, embedded credentials, or regulated data. Because the skill's purpose is to transform operational materials into a new artifact, unrestricted ingestion can cause sensitive content to be retained, summarized, or reproduced in the resulting skill.
The file defines activation via phrases such as "技能评估 / skill评估 / 评估技能," but it does not explain whether these are the complete trigger list, how exact matching works, or what nearby phrasings should not activate the skill. For a markdown file, this can create ambiguity around invocation scope.
The example recommendation tells users to invoke skills with Chinese phrases only, and similar user-facing instructions elsewhere in the document are also Chinese-centric. Because the file does not offer an opt-in language choice or explain that the skill is intended only for Chinese-speaking users, this can violate the language/locale policy requirement.
The file treats a phrase containing "做个skill" with little context as sufficient to route into the adaptive interview flow. While somewhat domain-related, it still lacks explicit constraints on what kinds of requests qualify and does not include negative examples for nearby but non-matching requests.
This file contains user-facing guidance and examples in both Chinese and English, but it does not state whether the skill should respond in Chinese, English, or adapt to the user's preference. Under the language/locale policy, forcing or implicitly assuming a language without opt-in can be a natural-language policy issue.
No suspicious patterns detected.