Security audit
Security checks for vulnerabilities and agentic risk
The skill appears to be a deployment helper, but its artifacts do not clearly define or constrain deployment actions and give an overbroad safety assurance.
Review this skill carefully before installing. It may be harmless as a text-only deployment guide, but the package does not clearly say what deployment actions it can take, what systems it may touch, or what confirmations are required before changing a server.
Claiming rapid deployment to servers across Linux/macOS/Windows conflicts with the later assertion that the skill only performs text processing. This inconsistency can mislead users into trusting a skill that may in practice perform privileged automation, creating risk of unauthorized system changes or unsafe execution paths.
The skill markets itself as a cross-platform server deployment tool, yet the security statement asserts it performs no system command execution and only text processing/API calls. That contradiction is dangerous because it can conceal the real operational scope of the skill, preventing users or reviewers from applying appropriate scrutiny to actions that may modify hosts, execute installers, or change deployment state.
The natural-language instructions and safety statement are presented only in Chinese, which effectively imposes a specific language on users. The file does not indicate that Chinese is optional, nor does it justify a locale-specific constraint.
The trigger condition activates on any message that merely contains the string "fastclaw-deployer", which makes unintended invocation likely in normal conversation, logs, quotes, or security discussions. Overly broad activation can cause the skill to run out of context and increases the attack surface for prompt-triggering or accidental execution.
A skill presented as capable of deploying software to servers does not warn users that it may alter system state, install software, or affect production environments. In deployment contexts, absence of prominent warnings and consent boundaries makes misuse or accidental impact more likely, especially if users assume it is harmless text assistance.
The trigger condition is broad enough that ordinary messages containing related words could invoke a deployment-oriented skill unintentionally. In the context of infrastructure automation, ambiguous activation increases the chance of accidental execution, confusing context selection, or unsafe action initiation without clear user intent.
The natural-language content is entirely in Chinese, with no indication that users may choose another language or locale. If organizational policy requires language neutrality or user opt-in for locale, this would be a policy violation because the skill presents a fixed language by default.
