Security audit
Security checks for vulnerabilities and agentic risk
This is a straightforward Chinese WeChat Moments copywriting skill that calls DeepSeek to generate marketing text, with no evidence of hidden persistence, destructive behavior, or unrelated access.
Install only if you are comfortable sending the product details you provide to DeepSeek and storing or exposing a DeepSeek API key through the documented environment or OpenClaw config path. Avoid entering confidential business plans, unreleased product details, or personal data unless that third-party API use is acceptable for your workflow.
The trigger phrase is very broad because it is framed like a natural-language request a user might say in ordinary conversation. In agent environments that auto-match skills from user utterances, this can cause unintended activation, routing normal chats into this skill, and creating prompt-surface expansion or incorrect tool use.
The skill name and description are entirely in Chinese and imply a Chinese-language output experience, but there is no indication that users can choose another language or that the locale restriction is intentional and justified. This can violate language or locale policy when the skill appears in a broader multilingual environment without opt-in.
The manifest description says the skill can 'one-click generate high-quality Moments marketing copy' for broad commercial use, but it provides no specific activation phrases, scope limits, or exclusion conditions. For a manifest file, this kind of generic invocation description is ambiguous and could overlap with common requests for writing help or marketing text generation.
The file description and prompt strings indicate the skill is specifically a Chinese-language social-post generator, and the system prompt instructs the model entirely in Chinese without any user opt-in or locale selection. This can violate language/locale policy when a skill imposes a specific language by default without offering choice or clearly documenting a justified regional constraint.
The skill sends user-supplied product text to DeepSeek over the network without any notice, consent flow, or minimization. If users provide sensitive business details, personal data, or confidential product information, that data is disclosed to a third-party service unexpectedly, creating a privacy and data-governance risk.
The natural-language prompt sent to the model is written to produce Chinese朋友圈 marketing copy and does not provide any mechanism for user language preference. Because locale is effectively forced by the instruction set, this is a natural-language policy concern unless the restriction is explicitly justified and documented.
The skill is entirely described and exemplified in Chinese, with no indication that users may choose another language or that the locale restriction is intentional and justified. This can conflict with language/locale policy expectations when a skill implicitly assumes a single language without opt-in.
The code accesses a sensitive environment variable and a local config file to obtain an API key, but provides no visible disclosure or explanatory comment about credential use beyond a minimal internal note. For safety review, sensitive credential access should be accompanied by clear user-facing documentation or warning unless already obvious from the skill's documented purpose.
No suspicious patterns detected.
