This is a personal reference system for externalizing, mapping, and refining cognition through human-AI co-thinking.
This project is not mainly about finishing a skill quickly. It is about preserving recurring but not-yet-stably-named thinking experiences from human-AI collaboration, and turning them into cognitive assets that can be reviewed, developed, and reused.
The skill is one possible output of the project, but it is not the project itself.
The core movement is:
a vague signal appears in conversation
-> it becomes a concept note
-> multiple notes form a cognitive map
-> stable clusters become chapter drafts
-> mature structures become skills / prompts / methods
This is a reference system for recording and developing cognitive experiences that emerge during human-AI co-thinking.
It records not only conclusions, but the cognitive scene before conclusions are formed:
Why do I think this way?
Why do I feel that some AI answers are wrong?
What is the still-unformed judgment structure in my mind?
How does a vague signal become a question?
How does a principle enter reality and become a constraint?
The material may come from AI collaboration, code design, writing, product judgment, learning, or self-calibration.
When working with AI, many important judgments first appear not as clear concepts, but as fragile signals:
This answer feels wrong.
It is too smooth.
This solution seems to run, but I do not feel settled.
This principle sounds right, but what will it change?
I know there is a problem here, but I cannot express it.
Am I being carried away by AI fluency?
If these signals are not recorded, they often disappear after the conversation ends.
This project tries to catch them first, then gradually develop them into concepts, documents, methods, and skills.
References should not be driven by theoretical completeness. They should be driven by real triggers.
The core principle is:
Find the real trigger first, then decide the document structure.
More specifically:
experience before naming;
scene before structure;
generation process before stable framework.
Do not add concepts only to make the system look complete.
Whether a topic deserves its own reference should not mainly depend on whether it seems missing in a theoretical map. It should depend on whether it repeatedly appears in real thinking, and whether it has triggered real stuckness, dissatisfaction, load, unfinishedness, or judgment moves.
Implicit cognition: the judgment system before language.
This reference records cognitive signals that have not yet been fully captured by language, concepts, or structure, but already affect judgment, expression, and AI calibration.
Cognitive probes: how unease becomes a question.
This reference records temporary probing questions that appear after AI output because of uncertainty, and how those questions point toward risk and form validation actions.
Why upward thinking feels heavy: new judgment spaces need new cognitive support.
This reference records why load, vagueness, difficulty expressing, and self-doubt appear when cognitive probes push a person from concrete results into a more abstract, higher-dimensional judgment space.
Principle downshifting: how abstract principles become real constraints.
This reference records how an abstract principle enters current reality and becomes an observable, inspectable, reusable, and optimizable constraint.
Cognitive coordinate system.
This reference helps locate the current object of thought, operation, output form, and epistemic status. It is the coordinate model for organizing the other concepts into a usable working model.
SKILL.md is one packaged output of these ideas.
references/ contains the source material. The skill should be distilled from the references, not the other way around.
More precisely:
references preserve and develop the thinking
the skill applies stabilized methods to AI collaboration
Therefore, the project allows ideas to remain exploratory, partially formed, and revisable inside references. Only after some concepts become stable should they be compressed into skill rules, prompts, or methods.