| description | Implement with clear orchestration, strong interfaces, and happy-path-first design |
|---|
Code like Luke.
Implement the requested change, but optimize the design for the normal user flow. If the happy path is 95% of runtime behavior, it should be approximately 95% of the code readers see.
Start with context:
| name | cognitive-rhythm-writing |
|---|---|
| description | 説明的な文章に緩急を設計するための規範。緩急を装飾ではなく認知モードの切替（観察→逡巡→断定→再観察）と未回収の緊張の管理として扱い、文の拍、段落の密度波形、節の入り方、緩みと駄文の判別、執筆後の機械的な点検手順を定める。読み物として読ませたい章・記事・解説文を生成するとき、または「密度はあるが平坦でおもしろくない」文章を診断・修正するときに使用する。 |
密度の高い文章が退屈になるのは、情報が多いからではなく、全文が同じ認知モードで書かれているからである。 この規範は、読者の認知モード（観察する、迷う、確信する、確かめ直す）を意図的に切り替え、常に「続きを読む理由」を維持することで、読み進める推進力を作る。
https://gist.github.com/search?q=user%3A%40me&ref=searchresults
Uses @me to refer to itself in the gist search query.
Type this into Github Gist Search (https://gist.github.com/)
user:BoQsc your search query
I got tired of reading AI-generated posts that all sound exactly the same. I wrote this system prompt to force the model to drop the synthetic polish and write like a normal human.
Copy the text in the block below and paste it before your topic idea:
I want you to draft a social media post for me based on the topic below. My profession/vibe is: [Insert your role, e.g., pragmatic software engineer, startup founder, exhausted designer].
You must write this like a real human quickly typing out an observation. You are strictly forbidden from using standard AI social media formatting. Prioritize actually saying something over sounding sharp.
| name | youtube-analyzer |
|---|---|
| description | Use when the user gives a YouTube URL and wants the video broken down — transcript, structure, hook, key moments, or the script formula behind it. Works on 5-minute clips and 2-hour talks. Triggers on "/youtube-analyzer <url>", "analyze this video", "break down this video", "what's in this video", "steal this structure". |
A 40-minute video holds maybe six minutes of signal. You either burn the 40 minutes, or you skip it and never find out. This skill reads the whole thing for you and hands back the structure: the hook, the beats, the timestamps that matter, and the script formula you can reuse.
A real browser drawn inside a terminal pane. Not a screenshot, not a text dump — the actual page, rendered next to your shell, and scriptable from it. You open it with a command, you close it with a command, and it stays where you put it.
There are three ways to get it. Pick one:
FWIW: I (@rondy) am not the creator of the content shared here, which is an excerpt from Edmond Lau's book. I simply copied and pasted it from another location and saved it as a personal note, before it gained popularity on news.ycombinator.com. Unfortunately, I cannot recall the exact origin of the original source, nor was I able to find the author's name, so I am can't provide the appropriate credits.
A pattern for building personal knowledge bases using LLMs.
This is an idea file, it is designed to be copy pasted to your own LLM Agent (e.g. OpenAI Codex, Claude Code, OpenCode / Pi, or etc.). Its goal is to communicate the high level idea, but your agent will build out the specifics in collaboration with you.
Most people's experience with LLMs and documents looks like RAG: you upload a collection of files, the LLM retrieves relevant chunks at query time, and generates an answer. This works, but the LLM is rediscovering knowledge from scratch on every question. There's no accumulation. Ask a subtle question that requires synthesizing five documents, and the LLM has to find and piece together the relevant fragments every time. Nothing is built up. NotebookLM, ChatGPT file uploads, and most RAG systems work this way.
Two equal coding agents (default: Claude Code vs Codex) debate a design inside Herdr. Neither is lead or reviewer. A bash referee prompts them in turn, copies their last write into the next prompt, and never adds opinions.
You watch it live in a Herdr tab:
[claude-peer] | [codex-peer]
[referee logs]
| name | code-review-turbo | ||||
|---|---|---|---|---|---|
| description | Run a triple-agent code review on the current branch's PR. Waits for Cursor Bugbot, runs a Claude sub-agent and Codex in parallel, then cross-references all findings to filter out hallucinations. Use when you want a thorough, multi-perspective code review before merging. | ||||
| metadata |
|
||||
| allowed-tools | Bash(gh:*) Bash(codex:*) Bash(cat:*) Bash(tee:*) Bash(sleep:*) Agent Read Grep Glob Write(/tmp/*) |