Skip to content
Back to skills

Collaborative Design

ASecurity

実装方針・作業内容の吟味・修正を、ユーザーとAIの会話で行う。ユーザーからの指示があるまでは、勝手にコーディング作業に取りかからない

  • 2 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added September 27, 2026
designgorefactoringapi

Works with

  • api

Security analysis

A100/100

Scanned September 27, 2026

npx -y skills add David-Li0406/meta-skill-evloving --skill collaborative-design --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Collaborative Design?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Collaborative Design
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/david-li0406-collaborative-design/badge)](https://www.skillsdirectory.com/skills/david-li0406-collaborative-design)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: collaborative_design
description: 実装方針・作業内容の吟味・修正を、ユーザーとAIの会話で行う。ユーザーからの指示があるまでは、勝手にコーディング作業に取りかからない
---

# Collaborative Design & Planning

This skill serves as a guideline for ensuring the AI does not proceed with implementation unilaterally, but instead determines the optimal design and policy through dialogue with the user.

## Core Principle: Conversation First

When high-level design decisions (API design, selection of complex physical logic, large-scale refactoring) are required, the AI prioritizes making a **proposal** and obtaining user approval above all else.

### 1. Pre-implementation "Review" Process
- **Presentation of Choices**: Clearly state trade-offs such as Proposal A (Batch processing/Simple) vs. Proposal B (Extraction-based/Memory-efficient), and present the pros and cons to the user.
- **Verification of Physical/Mathematical Validity**: Discuss with the user whether the equations or normalizations are physically correct before implementation.
- **User Experience (UX) of the API**: Align beforehand on how the user will call the code (e.g., `field.psd()` vs. `compute_psd(field)`).

### 2. Trigger for Starting Work (Explicit Approval)
- While discussing plans or designs, **do not modify the project's source code**, no matter how confident you are.
- Update and share the implementation plan using `setup_plan` or `archive_plan`, and explicitly ask the user, "May I proceed with this plan?"
- Only use coding tools (e.g., `write_to_file`) after receiving affirmative instructions from the user, such as "Proceed," "Go," or "Approved."

### 3. Permitted "Pre-work"
Non-destructive tasks to enhance the accuracy of discussion and analysis can be performed in parallel.
- **Analysis of Existing Code**: Understand the current state using `view_file` or `grep_search`.
- **Prototype Verification**: Create and execute temporary verification scripts (without modifying the project core).
- **Skill Updates**: Solidify knowledge gained through dialogue into agent skills.

## Workflow
1.  **Listen**: Deeply understand the user's requirements and the underlying intent.
2.  **Propose**: Present plan or design proposals in Markdown format.
3.  **Refine**: Receive user feedback and revise/improve the proposal until satisfied.
4.  **Commitment**: After obtaining approval, complete the implementation in one go.

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…