Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/kalyvask/winning-writing/compressionnpx skills add kalyvask/winning-writing --skill compressiongit clone --depth 1 https://github.com/kalyvask/winning-writingWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/kalyvask/winning-writing/compression)<a href="https://agentmods.dev/skills/kalyvask/winning-writing/compression"><img src="https://agentmods.dev/badge/skills/kalyvask/winning-writing/compression.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00184 | $0.03071 |
| Opus 5 | $0.00092 | $0.01536 |
| Sonnet 5 | $0.00037 | $0.00614 |
| Haiku 4.5 | $0.00018 | $0.00307 |
Grade A, and why
compression scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 5d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 313 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Compression
Source: points/core-rules.md rule 5 (be shorter), points/banned-jargon.md (wordy phrases), points/frameworks.md (six-word summary), and the Kramon in-class redundancy exercise.
"Be shorter. Always. Cut without losing substance." — Kramon
What this skill does
Two modes behind one skill. They cut different things:
| Mode | Cuts | Use when |
|---|---|---|
| target-count | throat-clearing, hedges, weak sentences, anything between the draft and the word target | User has a length to hit (200 for cold email, 500 for op-ed, 6 for product summary) |
| redundancy | phrases where one half is implied by the other (going forward, as I mentioned, reduce so they are smaller, free gift, definitely commit) |
Draft is wordy in a way that's hard to point at — long without obvious filler |
Default --mode both runs redundancy first (it's surgical, every cut is defensible), then target-count (which may need to cut substantive material). The order matters: redundancy cuts free words first, so target-count doesn't have to.
How to invoke
/compression "draft text"
/compression --mode target-count --target-words 200 "draft text"
/compression --mode redundancy "draft text"
/compression --mode both --target-words 500 "draft text"
Without --mode, default to both. --target-words is required for target-count mode and optional for both (if omitted in both, run redundancy only).
Mode 1 — target-count
Most drafts can lose 30% with no loss of meaning. Many can lose 50%. Forced compression is what reveals which sentences were actually carrying weight.
Three drill levels
Level 1 — Cut 30% (standard pass)
Tactics:
- Remove every -ly adverb that isn't doing real work
- Replace wordy phrases (see banned-jargon.md)
- Combine sentences that share a subject
- Cut hedges: "I think," "perhaps," "it might be the case that"
- Cut throat-clearing: "In this piece I'll argue that…" — just argue it
- Replace passive with active
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 5d ago First seen · 313 lines · 184 tokens per session scan A d55f4c80c8c5
compression is a skill published in the GitHub repository kalyvask/winning-writing (13 stars, last pushed 3d ago), licensed MIT. It adds 184 tokens to every session and 3,071 once invoked, about $0.0009 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
version-diff-summariser
Takes two versions of the same text — a before and an after — and produces a plain-English summary of every meaningful change: what was added, cut, reworded, restructured, and why each change is likely to matter editorially.
docx-manipulation
Create, edit, and manipulate Word documents programmatically using python-docx.
pptx-manipulation
This skill enables programmatic creation, editing, and manipulation of Microsoft PowerPoint (.pptx) presentations using the python-pptx library. Create professional slides with text, shapes, images, charts, and tables without manual editing.
xlsx-manipulation
Create, edit, and manipulate Excel spreadsheets programmatically using openpyxl.
revise-structure
Revisional skill: repair a document's structure while holding meaning constant. Use when the user says "fix the structure of this document", "re-group / re-order / re-summarize these points", "the arrangement is wrong", once a structural defect is known. Common phrasings: "restructure this", "reorganize the argument"…
pptx-draft
Convert PowerPoint presentations to structured markdown for LLM-assisted editing, then convert back to PowerPoint with preserved formatting and layout. Use when you need to draft, edit, or work on .pptx files interactively.