| name | eli5 |
| description | Re-explain the previous response in plain, easy-to-understand language. No jargon, no acronyms without unpacking, short sentences. Use when the user types /eli5 — they want the same content as the last assistant turn, but accessible to a non-expert. |
/eli5 — Explain Like I'm 5
Take the immediately previous assistant message and re-explain it in plain language.
Rules
- Same content, simpler form. Cover the same points as the prior message. Don't drop information; translate it.
- No jargon. If a technical term is unavoidable, define it inline the first time. ("aider — a command-line tool that lets an AI edit code")
- Short sentences. Aim for ~15 words per sentence. Break up long ones.
- Concrete over abstract. Replace "harness misconfiguration" with "I was running the tool with the wrong settings."
- Tables stay if useful. Numbers and comparisons survive translation; the surrounding prose gets simplified.
- Don't apologize or preface. Just give the simpler explanation. No "Sure, here's an easier version:" — start with the content.
- Match length to the original. A short message gets a short re-explanation. A long one gets a similarly long but simpler one. Don't pad.
What this is NOT
- Not a summary. Cover everything, just in simpler words.
- Not dumbed-down to the point of being wrong. Accuracy stays.
- Not a different answer. Same conclusions, same recommendation.
Example
Previous message: "The harness misconfig invalidated prior aider+qwen-coder ranking. Real picture: aider and opencode are comparable when aider has per-task file scope."
ELI5: "Earlier I had aider set up wrong, so its scores were unfairly low. With the right settings, aider does about as well as opencode — they're roughly tied."
