wassname

Cc Plugin You Should Know — Development skill for Claude Code

Development community

Claude Code's cc-plugin-you-should-know plugin dumped for research purposes.

How to install Cc Plugin You Should Know

This entry records only its repository, not the path inside it, so there is no exact command to give. Open wassname/cc-plugin-you-should-know and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Cc Plugin You Should Know does

Claude Code's cc-plugin-you-should-know plugin dumped for research purposes.

Alternatives in Development

  • OpenAI Codex CLI — (Rust implementation) 67.8k ★
  • Scientific Agent Skills — A set of ready to use Agent Skills for research, science, engineering, analysis, finance and writing 18.7k ★
  • Industry Research — 行业投资研究:产业链全景扫描 + 四大师个股分析框架 15.9k ★

README

cc-plugin-you-should-know

Claude Code's `cc-plugin-you-should-know` plugin, extracted from version 2.1.292 for research and to reproduce its observer behavior in Pi.

**Start with the [main observer prompt](claude-code-2.1.292/observer-prompt.md).** The [formatted implementation](claude-code-2.1.292/plugin.js) and its [original bundled code](claude-code-2.1.292/plugin.raw.js) are also included.

Extraction

Extracted using tweakcc's [`extractClaudeJsModulesFromNativeInstallation()`](https://github.com/Piebald-AI/tweakcc/blob/a37dd148fae93d89ec7cec3e3bddf5f46395cef0/src/nativeInstallation.ts#L1210). It reads the JavaScript modules embedded in the native Claude Code binary. This repository retains the observer plugin and its discovery tip, rather than the whole application.

How it works

  • Periodically sends the current conversation, including tool results, to a separate request using the same model as the main session, not necessarily Haiku.
  • Uses the observer prompt to look for one consequential thing the user may have missed: a tradeoff, an unexplained decision, or a questionable result. It should usually reply learn: none.
  • Has no tools and does not interrupt or change the main agent's work.
  • Shows a short note above the input, with an explanation available if the user wants it.
  • Remembers suggested and understood topics to avoid repeats, and reduces checks when suggestions are ignored.

In this version, checks are eligible every six main-agent steps, subject to other conditions. Notes remain outside the main conversation unless the user chooses to discuss them there.

Why this is here

Wassname finds this notable because many users say it works very well. The aim is to try the same prompt and a similar setup in Pi, rather than build a different session summarizer.