tech-preferences

Purpose

Current Known Preferences (Baseline)

Python Ecosystem

• Package management: uv (replaces pip/poetry/pyenv)
• Code quality: ruff (lint + format), ty (type check)
• Pre-commit/CI hooks: prek (replaces pre-commit)
• Automatic dependency upgrade: Renovate
• Data model: dataclasses (preferred, native zero dependency); introduce Pydantic v2 only when runtime validation or serialization is required

Web Backend

• Framework: FastAPI (async-first, automatic OpenAPI documentation)
• Input validation/serialization (API boundary): Pydantic v2

Web Frontend

• Framework: Svelte (lightweight, compile-time optimization, no runtime vdom)
• Build tool: Vite

System / Automation

• Task running: just (replaces Makefile)
• Containers: Docker / OCI images
• macOS daemon: launchd (do not use cron)

Data Format

• Configuration: TOML (preferred), YAML (CI only), JSON (API exchange)
• Text format: Markdown (documentation), structured logs prefer JSON

Preference Analysis Process

1. Identify decision points — Clarify the technical dimension to be selected (language/framework/tool/data format, etc.).
2. Compare with current preferences — Check the "Known Preferences" list above to confirm if there is an existing agreement; follow it by default if any.
3. Analyze deviation reasons — If you need to deviate from known preferences, clearly write down the reasons (performance, ecosystem, constraints, etc.).
4. Discuss new preferences — If you find new preferences worth solidifying, clearly propose them in the response:

   It is recommended to add "X replaces Y, reason: ..." to the tech-preferences skill.

5. Initiate proposal independently — When there are obvious new additions or changes to preferences, actively submit a preference change proposal in PR/Issue (see format below) without waiting for human prompts.
6. Deposit and update — After confirmation by humans, append the new preferences to the "Current Known Preferences" section.

Example Analysis

1. Decision point: Testing framework selection
2. Known preference: None (the current baseline does not list testing frameworks)
3. Compare candidates:

   pytest

   (richest ecosystem, prek integrated by default) vs

   unittest

   (standard library, no extra dependencies)
4. Analysis: All existing repositories use

   pytest

   ;

   prek

   hook runs

   pytest

   by default; choose

   pytest

5. New preference proposal: pytest (preferred Python testing framework) → Submit proposal to add to the baseline

How to Contribute New Preferences

偏好提案：将 <新工具/框架> 加入 tech-preferences skill
- 类别：<Python生态 / Web后端 / 前端 / 系统 / 数据格式>
- 替代：<被替代的工具，若有>
- 原因：<具体理由，1-3句>
- 参考：<项目/PR/文档链接，若有>