Dedao Brain Membership Thinking

What problem it addresses

When AI writing tends to over-reach, add unstated opinions, or lean on generic filler, this skill package splits the task into six tightly scoped modes: polish, comment, interrogate, review, create, and sprout. It does not try to produce a broad one-shot answer. Instead, it first classifies whether the user wants smoothing, highlight review, critical challenge, pattern review, channel-ready content, or cross-domain association.

How it works

• Mode selection: choose the matching reference file based on the request, for example references/polish.md for smoothing and references/interrogate.md for finding gaps.
• Rule execution: each mode defines what to do, what to avoid, banned phrases, and quality standards. polish follows clear, fluent, accurate, and concise constraints without adding unprovided viewpoints; comment identifies specific highlights instead of generic praise; interrogate performs critical checks and requires alternative proposals.
• Shared discipline: distinguish first-person original content from sourced content, keep output proportional to the input material, check generated text against banned phrases, and do not invent missing data.

Boundary notes

Useful for constrained, reviewable content workflows such as copy polishing, plan critique, record review, and channel drafting. If the request is ambiguous, ask which mode applies instead of guessing. In the provided materials, review, create, and sprout rely on user-supplied notes or files and should not assume access to external data sources.