latent-sre/homelab-agents/code-craft

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Language-level engineering craft — the idioms, traps, and testing discipline of the language at hand, plus how to change working code safely. Use when writing or reviewing Python, TypeScript, Bash, PowerShell, or Go, when adding tests to existing code, or when refactoring something that already works. sde-agents:backend-craft owns the service, sde-agents:frontend-craft the UI, sde-agents:sde-fullstack the end-to-end feature — this owns the code inside those layers.

Install

npx vskill@latest install latent-sre/homelab-agents/code-craft

SKILL.md

Code craft

sde-agents:sde-fullstack requires you to match the codebase's idioms. This is where they live.

The repository outranks this file. Every rule here is the default when the codebase has no opinion; where the surrounding code has an established pattern, match the code and apply the reason behind the rule through its conventions. A file written in the house style is worth more than a file written in the best style.

Two rules that hold in every language:

  • The dangerous operations are the silent ones. A crash is a good outcome; the failure modes

worth memorizing are the ones that continue with wrong state — a swallowed exception, an unchecked error return, a loop that exits early with a zero status, a dry-run that dries the wrong half.

  • Read the neighbors before writing. Two files near what you're changing tell you more about

what will be accepted than any style guide.

Load the reference for what you're writing

If the work involves…Read first
Python`references/python.md`
TypeScript or JavaScript — language-level code in any framework`references/typescript.md`
Bash or a shell script`references/bash.md`
PowerShell`references/powershell.md`
Go`references/go.md`
writing tests, or code that has none`references/tdd.md`
changing code that already works`references/safe-refactor.md`

A language not listed is not an exemption: apply the two rules above, read the neighbors, and follow the ecosystem's own idiomatic guide (its standard formatter and linter settle style — argue with neither).

Trips two predicates? Read both. Writing new Python with tests and touching working code trips three, and all three apply.

The review packet is the end-of-task report defined by the calling agent. Invoked standalone with no packet convention in context, end with: Changed / Assumptions / Verified / Not verified — and name which references you read.

Security

  • Security scanPassed, 95 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

7-day trend score
5

Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/latent-sre/homelab-agents/code-craft/badge)](https://verified-skill.com/skills/latent-sre/homelab-agents/code-craft)

Something wrong with this skill or its page? Tell us.