knowhub

name

knowhub

description

Use when turning a user's domain keywords, concepts, lived scenes, opinions, mistakes, and practical insights into a durable cognitive project with methodology, coaching system, references, cases, and reusable skills. Also use when routing to existing domains under domains/, including emotional-relationship, writing-storycraft, ai-agent, business-model, philosophy-reality, and trading-review, or when packaging subscription skill/MCP products under products/.

KnowHub

Use this skill when the user wants to build a cognitive project, not just receive a one-off explanation.

Core premise:

A cognitive project is not AI summarizing knowledge.
It is the user's experience systematized into models, coaching flows, cases, and skills.

For the full method, read:

methodology/cognitive-project-methodology.zh-CN.md

Domain Routing

Existing domains live under:

domains/

If the user asks about emotional relationship analysis, attraction stages, signal reading, intimacy boundaries, dignified expression, or relationship repair, route to:

domains/emotional-relationship/SKILL_技能说明中文版.md

If the user asks about writing, literature as a way to study people, plot development, real-life story simulation, immersive perspective-taking, character fixation, subtext, scene writing, or story audit, route to:

domains/writing-storycraft/SKILL_技能说明中文版.md

If the user provides a real relationship story and wants both real-world guidance and narrative simulation, read the emotional relationship domain first, then the writing-storycraft domain.

If the user asks about AI/Agent domain content, beginner AI education, Agent architecture, daily "what I did with AI" notes, enterprise AI scenarios, personal AI side-hustle methods, skill/MCP engineering, or real project breakdowns, route to:

domains/ai-agent/SKILL_技能说明中文版.md

If the user asks about philosophy-flavored content, reality simulation, Earth Online framing, life dungeon metaphors, relationship state changes, Wang Yangming-style action checks, or sharp opinion content, route to:

domains/philosophy-reality/SKILL_技能说明中文版.md

If the user asks about trading review, AI-assisted trading review, trading journals, emotional discipline, mistake pattern libraries, or trading content, route to:

domains/trading-review/SKILL_技能说明中文版.md

If the user asks about the product matrix, pricing, subscription skill/MCP products, creator accounts, or private-site conversion, read:

products/README_项目自述中文版.md
products/pricing.zh-CN.md
products/account-matrix.zh-CN.md

If the user asks about "one account per domain", which domain an account maps to, the emotional relationship account, AI Agent account, philosophy reality account, or trading review account, read the neutral mapping:

domains/account-domain-map.zh-CN.md
domains/<domain-name>/README_项目自述与概述.md

For product handoff, pricing, conversion, or account monetization, use the business model and products layer instead.

If the user asks about the business model, selling each paid skill separately, ebook-like purchase flows, account-bound entitlements, rich points, community Q&A incentives, sharing/referral loops, long-running point tasks, mini programs, WeChat public accounts, Toutiao, Zhihu, Xiaohongshu traffic, or website operations, route to:

domains/business-model/SKILL_技能说明中文版.md

Workflow

1. Collect seeds. Domain name, keywords, core concepts, lived scenes, mistakes, beliefs, target use.
2. Extract high-energy fragments. Repeated words, pain points, contradictions, failure modes, one-line insights.
3. Stabilize names. Turn vague feelings into reusable terms.
4. Build models. State machine, stage model, signal table, formula, red lines, checkers.
5. Make it coachable. Diagnosis order, evidence reading, strengths, risks, next action, scripts, review metrics.
6. Case it. Anonymize real scenes into repeatable training cases.
7. Projectize. Produce README, SKILL, methodology, references, cases, and domain index entries.
8. Iterate. Every new scene should validate, refine, split, or add modules.

Output Shape

Domain seed:
Keywords / concepts:
High-signal principles:
Core models:
Coaching flow:
Case list:
Project structure:
Next sampling questions:

Guardrails

• Do not write encyclopedia content when the user has not provided domain seeds.
• Do not flatten the user's sharp language into generic advice.
• Do not stop at quotes or feelings; turn them into models, coaching, and cases.
• Do not store raw private logs or identifiable third-party details.
• Do not declare the project complete before cases test the model.