deslop

Deslop

Remove the recognisable machinery of LLM prose from a piece of writing without changing what it says.

Two failure modes to avoid, in order of severity:

1. Changing the meaning. The point is to delete the performance, not the content. If a slop pattern is wrapped around a real claim, keep the claim and drop the wrapper.
2. Trading one tic for another. A rewrite that swaps "here's the thing" for "the reality is" has done nothing. Prefer deletion over substitution.

Before drafting

The rest of this skill cleans up a finished draft. When you're the one drafting a PR description, a docs page or a blog post, these keep most slop out of the first draft:

• Ask for a sample of the author's writing, or a piece they want to sound like, and match it. Models imitate an example far more reliably than they follow adjectives like "concise" or "natural".
• Name the reader before writing ("maintains a Panel app", "has never used HoloViews"). The reader decides what needs explaining.
• Describe the voice you want instead of listing tics to avoid. Naming a tic tends to make it show up.

The HoloViz voice, which the pr-description, documentation and outreach skills build on:

• Write in the first person, as the person who did the work, including what went wrong.
• Join cause and effect in one sentence with the connective that relates them (since, so, but) instead of a run of short beats, and vary the length. Joining can trade a [choppy-run] for a [justification-tail] once every sentence ends the same way, and a rescan catches that.
• Make each point once, and keep the concrete anchors (identifiers, numbers, a minimal example) while compressing.
• Use the word you'd say out loud to a colleague.

Workflow

1. Identify the target. A file path, a range in a file, or a draft sitting in the conversation. If the user says "deslop this" with no target, use the most recent substantial prose in the conversation. If the target is ambiguous between two candidates, ask.