skills-filter

---

name: skills-filter description: Evaluate whether a proposed skill is worth building by scoring it against five moat-oriented questions: differentiation, vertical fit, learning loop, user difficulty, and defensibility. Use when someone proposes a new skill and you want a fast build / deepen / reject decision before implementation. version: 1.0.0 author: Vivi tags: [skills, evaluation, strategy, triage] platforms: [all] dependencies:

• python3

---

skills-filter

Fast gate for new skill ideas. It prevents the team from spending time on generic wrappers and pushes skill design toward domain knowledge, data loops, workflow integration, and defensibility.

Use this skill when

• Someone says "我想做一个 XXX skill"
• You need to decide whether a skill idea deserves engineering time
• A proposed skill feels too generic and needs a harder filter
• The team wants a shared rubric before adding new items to MyStableSkills

Workflow

1. Ask the five questions.
2. Score the candidate out of 5.
3. If the score is 5/5, build it now.
4. If the score is 3-4/5, deepen the moat first.
5. If the score is 0-2/5, reject it or redesign it.

Commands

Show the checklist:

python3 main.py --template

Run a built-in example:

python3 main.py --example lawyer-email-assistant

Score a real proposal:

python3 main.py --title "创业者精力风控系统" --answers y,y,y,y,y

Questions

1. 跟"直接丢给 AI"有什么明显区别？
2. 有行业偏向性或场景偏向性吗？
3. 会越用越聪明吗？
4. 用户自己很难做到吗？
5. 有壁垒吗？

Inputs and outputs

• Input: candidate skill idea plus five yes/no judgments
• Output: per-question pass/fail, score out of 5, and a build/deepen/reject decision

Validation

Minimum validation commands:

python3 main.py --example lawyer-email-assistant

python3 -m py_compile main.py

Constraints

• This is a triage filter, not a replacement for later product design
• A high score means "worth building", not "already well specified"
• Use it before implementation, not after the team has already sunk the cost