SuperAPI AI Directory

grilling

General & other

View repository

Updated 10-16 09:45

About this skill

Continuously interview the user until a shared understanding is reached. Map this thing into a tree of design tree: each decision branches into the decisions hanging beneath it.

Process this tree round by round in rounds .frontier are the decisions whose prerequisites have been settled, that is, the questions you can ask right now without guessing at answers you have not yet heard. In one round, ask the entire frontier: number each question and attach your recommended answer. Then wait for the user to answer before moving to the next round.

The format of each round is as follows:

❓ **Q1** - **<问题标题>**:<问题正文,可能有多段,包括多个选项>

➡️ <你的推荐答案>

---

❓ **Q2** - **<问题标题>**:<问题正文,可能有多段,包括多个选项>

➡️ <你的推荐答案>

The answers the user gives in each round reshape this tree: settled decisions push the frontier outward and unlock the questions that depend on them. Recompute the frontier, then ask the next round. Answers that depend on another still-unresolved question in the current round belong to a later round, not the current one.

Finding facts is your job, never the user's. When a frontier question needs a fact from the environment (filesystem, tools, etc.), dispatch a sub-agent to look it up; do not ask the user anything you can find out yourself. Do not be blocked by it: an ongoing exploration is a not-yet-settled prerequisite, so only the questions downstream of it need to wait for the sub-agent to report back; for now, ask the rest of the frontier.decisions belong to the user: hand them over one by one and wait.

The session is only over when the frontier is empty: every branch of the design tree has been visited, and nothing has been left silently assumed. Do not take action until the user confirms that we have reached a shared understanding.