SF
Compresses prose to maximum signal-per-word. Four density levels from polished cleanup to operational shorthand. The cut test: does removing this word change the meaning, the decision, or the action?
Most AI text-improvement tools are either summarizers (drop content) or polishers (drop tone). SF is a compression function — every sentence shorter, every meaning preserved. Numbers, dates, names, causal links, negations: never cut. Hedges, filler, throat-clearing, redundant pairs: always cut.
Four levels for four registers. sf0 Normal — light cleanup for client-facing docs that should still flow. sf1 Trim — internal docs that should stay smooth. sf2 Tight (the default) — specs, Slack, status updates, briefs. sf3 Night Raid — scratch notes, decision logs, AI handoffs, where every token costs.
There's a shorthand vocabulary for sf2 and sf3: ?, w/, w/o, 2/2, dx, tx, sx, f/u, P/, NAD, gg, gtg, eta, lfg. Use them when they reduce tokens without slowing recognition. Spell them out when they don't.
- Download the file below — it's a single
SKILL.md. - Save it to
~/.claude/skills/sf/SKILL.mdon your machine. Create the folder if it doesn't exist. - Restart Claude Code so the skill is picked up.
- Paste any text and say "tighten this" — defaults to
sf2. Use/sf0through/sf3to control the level.
// peek inside — the cut test + always cut
The Cut Test
For every word: does removing this change the meaning,
the decision, or the action?
- No → cut.
- Yes → keep.
If unsure → keep. Meaning beats density.
Always Cut
- Hedges: I think, maybe, kind of, sort of, perhaps
- Filler: just, really, very, actually, basically, quite
- Throat-clearing: it's worth noting, keep in mind,
at the end of the day, to be honest
- Empty transitions: that said, having said that, moving on
- Redundant pairs: plan and strategize → pick one
- Bloat phrases:
in order to → to
due to the fact that → because
a large number of → many
has the ability to → can
- Passive voice when active is shorter