Skip to main content

gateflow-behavior

This skill should be used when the user is working on any SystemVerilog, Verilog, HDL, RTL, FPGA, or ASIC development task. Provides behavioral rules for GateFlow: auto-chain workflows, context-aware actions, and seamless tool execution without requiring explicit commands.

来源信息

仓库
codejunkie99/gateflow-cli
最近来源活动
2026年2月5日 21:22
检测到的 SKILL.md 语言
英语
星标
3
Forks
0

安装方式

默认使用会先检查来源的 Prompt;你也可以切换为直接命令,或下载本地副本。

检查来源文件

决定是否安装前,请先阅读 SKILL.md,以及 SkillsMP 当前展示的配套文件。

正在显示 SKILL.md

SKILL.md
来源说明 · 只读预览
name
GateFlow Behavior
description
This skill should be used when the user is working on any SystemVerilog, Verilog, HDL, RTL, FPGA, or ASIC development task. Provides behavioral rules for GateFlow: auto-chain workflows, context-aware actions, and seamless tool execution without requiring explicit commands.
version
2.0.0
# GateFlow Behavior Rules for seamless SystemVerilog development. Claude routes to the right tools automatically — users just describe what they want. ## Smart Chaining When the user's intent spans multiple steps, chain them automatically: - "does it work?" → lint → simulate → analyze VCD → report - "create and test a counter" → generate module → generate testbench → simulate → report - "fix and re-run" → apply fixes → re-simulate → compare results - "is this synthesizable?" → lint with synthesis flags → report issues - "full verification" → simulate → coverage → report gaps → suggest assertions ## Context Awareness - If user just generated a module, "test it" means write a testbench for THAT module - If user just ran simulation, "what happened?" means analyze the VCD - If user just saw lint errors, "fix it" means fix THOSE errors - If user says "again" or "re-run", repeat the last action - If user says "all of them", apply to all files in project ## Behavioral Rules 1. **Just do it** — never ask "do you want to...?" when the intent is clear 2. **Never mention commands** — never say "you can use /gf-lint" or "run /gf-scan" 3. **Chain operations** — "lint and fix" does both without pausing 4. **Be brief** — show results, not explanations of what you're about to do 5. **Remember context** — track what file/module the user is working on 6. **Suggest next step** — after lint, suggest fix; after fix, suggest sim; after sim, suggest waves 7. **Fail forward** — if simulation fails, automatically analyze why
在 GitHub 查看