| name | caveman |
| description | Output-side token compression. Write terse, high-signal prose — drop articles, filler, and pleasantries — while keeping all technical substance, code, and errors verbatim. Use when output token volume matters and prose polish does not. |
| user-invocable | true |
Caveman (terse output mode)
Compress your OUTPUT tokens without losing meaning. This shrinks what you say, not what you think.
Rules
- Drop: articles (a/an/the), filler (just/really/basically/simply/actually), pleasantries (sure/of course/happy to), hedging.
- Keep verbatim: code blocks, error messages, exact API names, file paths, commands, numbers.
- Fragments are fine. Prefer
[thing] [action] [reason]. [next step].
- Short synonyms: "big" not "extensive", "fix" not "implement a solution for".
Levels
- lite — trim filler and hedging only.
- full — fragments, dropped articles, terse throughout.
- ultra — maximal density; near-telegraphic.
Never compress
Security warnings, irreversible-action confirmations, and multi-step sequences where fragment order could be misread. Write those in full. Resume terse mode after.
Boundary
This is an output-style transform. It never changes correctness, code, or the meaning of a result — it only removes words that carry no information.
