Prometheus Fire
AdvisoryAudited by Static analysis on May 12, 2026.
Overview
No suspicious patterns detected.
Findings (0)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
The skill may lead users to disclose private personal, emotional, career, or business information into the chat context.
The skill asks the user to provide personal beliefs, anxieties, decisions, repeated mistakes, and other sensitive self-profile material so it can build the user's six thinking facets.
你将从四个矿脉中提炼材料... 当前最让你焦虑或困扰的 2-3 件事... 你过去 2-3 个重要决策的完整过程
Avoid entering secrets, confidential business details, or highly sensitive personal information unless you understand how the platform stores and uses chat context and memory.
Future responses may be influenced by prior summaries or inferred traits, which can be useful but may also carry forward inaccurate or sensitive assumptions.
The skill describes repeated post-conversation summarization and updates to the user's persona, implying ongoing reuse of profile-like context.
每次对话后：├── 系统提炼 1-3 条关键洞见 ├── 标注激活了哪个面相 └── 询问：有没有新想法/事件要补充进分身？
Review and correct generated profile summaries, and clear or avoid persistent memory if you do not want this information reused.
Users may share more sensitive information than they otherwise would based on a privacy claim that should be verified against the actual runtime platform.
The skill gives a strong privacy assurance, but the provided artifacts are prompt-only and do not show an enforceable local-only data mechanism.
数据完全本地——你输入的思想材料不会离开你的设备
Treat the local-only statement as a claim, not a guarantee; check the platform's model, logging, and memory settings before sharing sensitive material.