Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed academic polishing tool that sends user-provided text to DeepSeek for rewriting, with no hidden persistence or unrelated local access found.
Install only if you are comfortable sending the text you submit to DeepSeek. Avoid using it for confidential, unpublished, regulated, or third-party-restricted manuscripts unless that external processing is permitted by your policies.
Skill contains instructions that could directly expose system prompts, internal rules, or hidden instructions to users or external parties.
),
},
}
return prompts.get(mode, prompts["polish"]).get(language, prompts["polish"]["zh"])
def polish_text(text: str, mode: str = "polish", language: str = "auto") -> dict:
The skill invokes external capabilities (environment access and network use via DeepSeek API) but does not declare any explicit tool scope or permissions boundaries. This creates an authorization and transparency gap: the runtime may permit broader capability use than users or reviewers expect, making misuse of secrets or outbound requests harder to audit and constrain.
The skill explicitly says user text will be sent to the DeepSeek API, but it provides no warning, consent step, or data-handling notice before transmitting potentially sensitive manuscript content to a third party. Academic drafts often contain unpublished research, personal data, or confidential material, so silent external transfer can cause privacy, confidentiality, and compliance issues.
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
# DeepSeek API 配置
DEEPSEEK_API_URL = "https://api.deepseek.com/v1/chat/completions"
SKILL_ID = 7665153655141580854
This request transmits user text and system instructions to an external service over the network. In the context of academic polishing, the text may contain confidential manuscripts, personal data, embargoed results, or proprietary research, making third-party transmission security-relevant even if TLS is used.
}
try:
response = requests.post(
DEEPSEEK_API_URL, headers=headers, json=payload, timeout=60
)
if response.status_code >= 400:
The skill sends the full user-supplied paper text to DeepSeek's external API, which can expose unpublished research, sensitive academic material, or regulated data to a third party. There is no in-code notice, consent flow, masking, or policy guardrail indicating that user content leaves the local environment.
The instruction requires Chinese replies for Chinese input and English replies for English input, imposing a language policy automatically. The file does not indicate that users may opt into a different reply language or override this behavior.
The code automatically selects either Chinese or English prompts and output behavior based on character heuristics when language is set to "auto". This imposes a language choice without explicit user opt-in at runtime, which can conflict with language/locale policy expectations requiring user choice.
No suspicious patterns detected.