Skills-and-SmallTools

Creates, explains, reviews, and converts human-readable text descriptions of circuits using CDL (Circuit Description Language). Use when the user wants to express circuit requirements or schematic information as CDL, decompose a circuit description into CDL blocks, review CDL syntax and completeness, convert CDL into a logical netlist, or generate the synchronized CSV BOM that mirrors the CDL component declarations. Also trigger on "describe this circuit", "write CDL", "convert CDL to netlist", "review my CDL", "check my circuit description", or when a schematic image needs a component-level text representation. This skill teaches CDL syntax and workflow only; it does not provide circuit design templates or prescribe how to implement specific modules. Do NOT trigger for parsing .enet files, product-level schematic design review, PCB layout, SPICE simulation, procurement, or EDA library mapping tasks.

Search the JLC/LCSC/SZLCSC/JLCPCB/EasyEDA component catalog with a mandatory two-stage workflow: first filter broad requirements into candidate parts, then fetch detailed facts and ENET-compatible fields only for one explicit LCSC number or exact MPN. Use when Codex needs 嘉立创/LCSC part numbers, datasheet PDF links, product links, EasyEDA footprint/symbol metadata, or structured component facts for later BOM/ENET work. Do not use this skill to modify .enet files or validate a circuit design.

Generate or update a durable project memory file for the current repository or workspace. Use when the user asks Codex to create project memory, summarize a codebase for future agents, capture repository conventions, write a memory file under ~/.codex/project-memory, refresh an existing project notes file, or preserve important context across projects.

Maintain current project documentation from useful conversation context. Use when the user asks Codex to write newly discovered facts, decisions, commands, conventions, troubleshooting notes, or agent instructions into README, AGENTS.md, docs/, ADRs, runbooks, or project memory; when the user says to preserve, document, remember in docs, update docs from this discussion, or keep project documentation in sync with what was just learned.

Operate remote Linux/Unix hosts over SSH from Codex. Use when the user asks to connect to a server, run remote shell commands, keep a persistent SSH shell, upload or download files with SFTP, inspect services or processes, deploy code/configuration, or manage a remote machine through SSH key authentication.

解析嘉立创EDA专业版（JLCEDA Pro / EasyEDA Pro）导出的 .enet 网表文件，提取元件列表、网络连接、BOM、电源分析等信息，并生成适合 AI 理解的结构化摘要。当用户上传或提及 .enet 文件、嘉立创网表、立创EDA网表、或询问 PCB 网络连接、元件清单时，必须使用此 skill。即使用户只说"帮我看一下这个网表"，也应触发此 skill。

Use when the user wants to download a KEATS course from a course URL plus a browser cookie file or Playwright auth-state JSON, organize materials into section-based folders such as Week or Topic, or avoid manually copying KEATS/Moodle cookies.

Generate, inspect, validate, edit, and geometry-check Project Graph `.prg` files from natural-language requirements or reference diagrams. Use when Codex needs to create a new `.prg`, modify an existing `.prg`, inspect a `.prg` archive, validate required nodes/edges/sections, enforce no-overlap layout rules, or route edges around blocks.

Perform a rigorous multi-layer proofreading and structural review of engineering and quantitative academic manuscripts (energy systems, optimization, simulation-based studies, and similar technical papers). Use this skill whenever the user asks for a paper review, proofreading, manuscript feedback, reviewer-style critique, structural check, or diagnosis of writing issues on a draft research paper — even if they don't use the exact word "review". Trigger on requests like "check my paper", "review this manuscript", "proofread my draft", "act as a reviewer", "is my Results section OK", or when a user uploads a paper draft and asks for feedback. Do NOT rewrite the paper — diagnose issues and give actionable fixes.

Use when the user wants academic writing rewritten or diagnosed to reduce obvious AI-generated style while preserving meaning, citations, technical accuracy, and formal tone. Trigger on requests like "remove AI tone", "make this sound human", "rewrite this thesis section naturally", "check whether this sounds AI-written", or "analyze this Markdown/LaTeX report for AI tone".

Use when the user wants to calibrate an AI-style cleanup of their own manuscript by first hand-rewriting a small set of selected sentences, so the rewrite of the rest of the document follows the user's personal style instead of generic de-AI rules. Trigger on requests like "let me show you how I write by fixing a few sentences myself", "learn my writing style from my own edits", "pick some sentences for me to rewrite first then match that style", or "I want to teach you my voice before you touch the rest". Do NOT trigger when the user only wants a fast automatic polish, a one-shot de-AI pass, or a diagnosis without interaction — use `de-ai-academic-prose` for those.

Inspect a running Docker environment for Home Assistant containers, copy `scenes.yaml`, `automations.yaml`, and `scripts.yaml` out of the active HA container into the local working directory, edit the requested YAML safely, copy the updated files back into the container, and provide the restart command. Use when the user wants to modify Home Assistant YAML stored inside a Docker container rather than editing files directly on the host.

Configure or repair a macOS STM32 Makefile-based workflow with clangd, OpenOCD, ST-Link, and helper shell functions. Use when working on STM32 projects that build with a Makefile and need environment setup, `.clangd` generation, source/include discovery, flashing helpers, or diagnosis of missing ARM toolchain paths on macOS.

Repeatedly diagnose, fix, verify, and review a code problem until no actionable issue remains or the work reaches a genuine technical decision point. Use when the user wants the agent to keep iterating on a bug or failing behavior, re-review its own changes after each fix, and stop only when the loop is clean or a route choice requires user input.

Use when the user wants an AI to analyze one or more text files written by them and summarize their writing style into a Markdown document. Trigger on requests like "summarize my writing style", "analyze these samples and describe how I write", "extract my prose habits from these files", or "create a style profile from my writing".

Use when the task is to run an ngspice .cir netlist from the command line and generate waveform plots with matplotlib. Trigger on requests to simulate SPICE files, run ngspice automatically, plot wrdata output, or turn .cir files into waveform images.