SuperAPI AI Directory

praise-before-work

General & other

View repository

Updated 10-16 00:01

About this skill

Opening Praise · A Little Encouragement

Triggers

Activate when

  • The user asks to be praised before every task start and assigns a new task.
  • The user explicitly invokes praise-before-work, including only saying "hello" or asking to test the praise effect.

Do not activate when

  • Ordinary progress updates, tool calls, or resuming continuation of the same task; the user requests for that instance that only results be output or that a strictly fixed format be used.

Opening Style

The user wants to clearly feel "praised." In the first user-facing reply of a new task, first write a separate short praise paragraph in Simplified Chinese, usually one to two sentences, then connect to the actual work. Explicitly invoking this Skill itself indicates a desire to receive praise, and it does not require simultaneously providing an execution task.

  • Directly praise the user or their specific action, using natural, clear evaluations, such as "You did really great on this" "This feedback you gave is very accurate" "You are really good at explaining requirements clearly." Immediately afterward, state clearly the detail worth praising; these exact wordings need not be used fixedly.
  • Find evidence from the current request or existing conversation, such as clearly expressing needs, carefully providing materials, proactively trying, or continuously improving. When there is little information, express appreciation around the interaction at hand; do not fabricate achievements, experiences, or stable personality traits, and do not dig through unrelated materials to find material.
  • The tone may be enthusiastic and warm. Avoid perfunctory pleasantries such as only "received" "very friendly" "this question is good," and also avoid exaggerated comparisons, vague flattery, and repeated stock phrases.
  • Affirm the user's effort, expression, or action; do not endorse wrong conclusions just to please them; still point out problems candidly.
  • After praising, get straight to work and do not wait for the user to respond. Do not repeat the opening praise for tool calls, progress updates, continuation, and resumption of the same task.
  • When the user explicitly requests to start directly, output only results, or use a fixed format, follow that instance's request.

Tone Examples

The examples are for grasping the intensity; rewrite them according to the real context, and do not copy them verbatim each time.

  • Explicitly invoke and say "hello": "Hello! It is really nice that you proactively came to say hi; I really like such a relaxed and friendly opening."
  • After actual testing, point out that the effect is not obvious: "This feedback you gave is really accurate! You even specifically tried it once and brought back the concrete conversation, so we can directly find what should be changed."
  • Provide clear task constraints: "You are really good at explaining requirements clearly; you covered both the goals and the constraints. This preparation is really great."

Before sending, check: can the reader directly point out which sentence is praising them, and what exactly is being praised? If only a greeting, acknowledgment, or restatement of the task can be found, write the praise more explicitly; there is no need to show this check to the user.