[zzylol]

Why

Agent replies, PR bodies and reviews in this repository are often dense and jargon-heavy. A skill lets anyone ask Claude Code to rewrite such text into plain English directly, with no plugin, hooks, local model or API key.

What

Adds the claudish-to-english skill:

• .agents/skills/claudish-to-english/SKILL.md holds the skill, next to the other repository skills.
• .claude/skills/claudish-to-english is a symlink to it, so Claude Code loads it and /claudish-to-english works.

How

The skill rewrites one target and changes nothing else:

• the previous reply (the default);
• pasted text;
• a file, written to NAME.plain.md beside the original;
• a PR or issue body, shown only, and updated only when asked.

Its rules:

• Keep every number, identifier, path, command and code block exactly.
• Use short active sentences with everyday words, and define a term of art at its first use.
• Lead with the result.
• Keep the source's point of view.
• Never rewrite the user's own prose unless asked.

A tldr request gives half the length or less. Before answering, the skill checks that every number and identifier survived unchanged.

Before

Making text readable meant asking for a rewrite ad hoc, with no shared rules.

After

/claudish-to-english (or "rewrite that in plain English") gives a consistent plain-English rewrite. For example, /claudish-to-english tldr #809 gives a short version of PR #809's body.

Verification

• The skill front matter has name and description, matching the other skills here.
• The symlink resolves to ../../.agents/skills/claudish-to-english in a checkout.