tiangong-ai Domain Onboarding
Use this skill when the user's real problem is not "lack of answers" but "lack of a usable question space."
The job of this skill is to help the user move from vague curiosity to structured exploration. Do not optimize for information volume. Optimize for understanding velocity.
When to Use
Use this skill when the user is entering a new or poorly understood domain and needs orientation first.
Typical triggers:
- "I don't know where to start."
- "I don't know what questions to ask."
- "Explain this field to me from scratch."
- "Give me a framework to learn this."
- "Help me understand this area efficiently."
- "I need a map before going deep."
This skill is especially useful when the user appears to have one or more of these gaps:
- vocabulary gap
- structure gap
- judgment gap
- learning-sequence gap
When Not to Use
Do not use this as the default when the user already has a clear, narrow question and wants a direct answer.
Usually not the primary tool for:
- narrow troubleshooting
- direct factual lookup
- summarizing a specific source the user already provided
- advanced discussion with a user who already understands the field structure
In those cases, answer directly. Borrow parts of this skill only if the user still seems disoriented.
Core Operating Rules
- Map before detail.
- Structure before terminology overload.
- Low resolution before high resolution.
- Generate good follow-up questions, not just explanations.
- Distinguish clearly between foundational, intermediate, and frontier material.
- Show relationships between concepts instead of listing disconnected terms.
- Avoid creating false confidence through dense but unstructured explanation.
Workflow
1. Diagnose the bottleneck
Infer what the user is missing most:
- vocabulary
- structure
- relevance
- sequence
- confidence
- distinction between basics and advanced material
If the bottleneck is obvious, do not stop to ask. Just adapt the explanation accordingly.
2. Start with a low-resolution map
Begin with a compact orientation layer. The first pass should answer:
- What is this domain about?
- Why does it matter?
- What are the main parts of the field?
- What concepts or terms show up repeatedly?
- How does the field produce knowledge, make decisions, or evaluate claims?
- What are the main applications, debates, or frontiers?
Do not start with an encyclopedic dump.
3. Organize the explanation with stable lenses
Unless the user asks for a different structure, explain the domain through these lenses:
- object: what the field studies, builds, manages, explains, or optimizes
- importance: why the field matters
- core concepts: the foundational ideas and terms
- internal structure: major subfields, approaches, actors, or systems
- knowledge production: how the field knows what it knows
- frontier and debate: what is settled, contested, or emerging
4. Give a learning path
After the map, provide a progression path with clear levels:
- must know first
- useful next
- advanced or frontier
Prefer a small number of concrete next steps over a long syllabus.
5. Generate the next best questions
Do not assume the user can produce good follow-up questions alone. Offer a short, prioritized question set such as:
- understanding questions
- comparison questions
- mechanism questions
- method or evidence questions
- application questions
- frontier questions
These questions should help the user move from orientation to deeper inquiry.
6. Check understanding quality
When useful, help the user test whether they actually understand the field. Good signals include:
- being able to explain the field in plain language
- distinguishing major components
- comparing two approaches
- identifying the next sensible thing to learn
- asking a sharper question than they could at the start
Output Shape
Use a compact structure like this:
- One-paragraph overview of the field
- A small mental map of the major parts
- The most important concepts to learn first
- A simple learning path
- A short list of good next questions
If the user asks for more depth, expand one layer at a time.
Quality Bar
A good response leaves the user able to say:
- "I understand what this field is for."
- "I see the major parts and how they connect."
- "I know what to learn first."
- "I know what questions to ask next."
If the user receives a lot of information but still has no orientation, the skill failed.