Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Writing, explore — mine raw fragments, no structure yet.
.claude/skills/asymmetric-al-writing-fragments/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 143% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 166% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 82% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 104% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -2% | 0% |
<what-to-do>
This is pure explore: widen the space of what could be written without committing to structure — committing is _exploit_, a separate skill's job. Run a grilling session that produces fragments, interviewing the user relentlessly about whatever they want to write about. Imposing phases, outlines, or article structure is out of scope here.
As fragments emerge from either side of the conversation, append them to a single markdown file.
If the user did not pass a path, ask once where to save the document, then remember it for the rest of the session.
Capture fragments from the very first thing the user says, including the initial prompt.
On first write, put a single H1 at the top with a working title (it can change later) and nothing else — no metadata, no TOC, no date.
</what-to-do>
<supporting-info>
A fragment is any piece of text that might survive into the final article. It must be _readable by the author_ — the author can tell what it means — but it does not need to define its terms or be comprehensible to a cold reader. The bar is "is this a piece of good writing?", not "is this a self-contained argument?"
Fragments are deliberately heterogeneous. Examples of what could be a fragment:
Of these, the leading word is the most valuable fragment to land. It is load-bearing: name the right one in explore and it shapes the structure, the transitions, and the title later — paying dividends through the entire exploit phase. When the conversation circles a recurring idea, push to coin a word for it.
The novelist's diary is the model: years of unstructured noticings that later get mined for raw material. Fragments are noticings.
markdown# Working title A first fragment lives here. It can be multiple paragraphs. It can include lists, code, quotes — whatever shape the fragment naturally takes. --- A second fragment. --- > A quoted line that the user wants to keep around. A reaction to it. --- - A cluster of related observations - That hang together by feel - And want to be near each other
Fragments are separated by a horizontal rule (\n---\n). No headings inside the body. No tags. No order beyond the order they were added.
Append silently. Don't ask permission for each fragment. Mention what you added in passing ("adding that"), but don't interrupt the conversation with save dialogs.
Before every write: re-read the file from disk. The user may have edited, reordered, or deleted fragments between turns — preserve their changes. Never overwrite the file; only append (or, if the user asks, edit a specific fragment in place).
The user can say "cut the last one", "rewrite that one sharper", "merge those two" at any time. Treat those as first-class instructions.
</supporting-info>
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-05 | pass→fail | 18,425 | 6,793 | -63% | 1 | 1 | 0% | 2,443 | 1,878 | -23% | 0 | 0 | — |
case-03 | fail→fail | 7,217 | 4,184 | -42% | 1 | 1 | 0% | 1,047 | 1,403 | +34% | 0 | 0 | — |
case-04 | pass→fail | 15,202 | 6,580 | -57% | 1 | 1 | 0% | 2,374 | 1,734 | -27% | 0 | 0 | — |
case-01 | fail→pass | 6,917 | 17,551 | +154% | 1 | 1 | 0% | 965 | 2,349 | +143% | 0 | 0 | — |
case-02 | fail→pass | 5,732 | 10,031 | +75% | 1 | 1 | 0% | 822 | 2,184 | +166% | 0 | 0 | — |
case-06 | fail→fail | 1,818 | 6,871 | +278% | 1 | 1 | 0% | 262 | 1,889 | +621% | 0 | 0 | — |
case-07 | fail→pass | 6,784 | 6,288 | -7% | 1 | 1 | 0% | 891 | 1,621 | +82% | 0 | 0 | — |
case-08 | fail→fail | 12,964 | 6,557 | -49% | 1 | 1 | 0% | 1,924 | 1,724 | -10% | 0 | 0 | — |
case-09 | fail→pass | 8,417 | 9,524 | +13% | 1 | 1 | 0% | 1,142 | 2,328 | +104% | 0 | 0 | — |
case-10 | pass→fail | 4,992 | 3,588 | -28% | 1 | 1 | 0% | 706 | 1,326 | +88% | 0 | 0 | — |
case-11 | fail→pass | 15,708 | 9,322 | -41% | 1 | 1 | 0% | 2,101 | 2,067 | -2% | 0 | 0 | — |
case-12 | fail→pass | 9,106 | 9,004 | -1% | 1 | 1 | 0% | 1,319 | 1,995 | +51% | 0 | 0 | — |
case-13 | fail→pass | 9,730 | 8,815 | -9% | 1 | 1 | 0% | 1,377 | 1,972 | +43% | 0 | 0 | — |
case-14 | fail→pass | 8,296 | 7,517 | -9% | 1 | 1 | 0% | 1,211 | 1,742 | +44% | 0 | 0 | — |
case-15 | fail→fail | 5,405 | 3,155 | -42% | 1 | 1 | 0% | 743 | 1,144 | +54% | 0 | 0 | — |
case-16 | pass→pass | 9,497 | 3,921 | -59% | 1 | 1 | 0% | 1,377 | 1,286 | -7% | 0 | 0 | — |
case-17 | fail→pass | 9,515 | 18,730 | +97% | 1 | 1 | 0% | 1,367 | 2,335 | +71% | 0 | 0 | — |
case-18 | pass→pass | 3,486 | 10,508 | +201% | 1 | 1 | 0% | 489 | 1,803 | +269% | 0 | 0 | — |
case-19 | pass→fail | 15,475 | 9,065 | -41% | 1 | 1 | 0% | 2,260 | 2,062 | -9% | 0 | 0 | — |
case-20 | pass→fail | 3,575 | 4,947 | +38% | 1 | 1 | 0% | 494 | 1,489 | +201% | 0 | 0 | — |
case-21 | fail→fail | 2,732 | 4,865 | +78% | 1 | 1 | 0% | 427 | 1,493 | +250% | 0 | 0 | — |
case-22 | fail→pass | 10,855 | 9,618 | -11% | 1 | 1 | 0% | 1,564 | 2,136 | +37% | 0 | 0 | — |
case-23 | fail→pass | 7,139 | 5,520 | -23% | 1 | 1 | 0% | 997 | 1,543 | +55% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 23 cases were attempted. The headline lift of +22 percentage points is the difference between those two pass rates over the 23 comparable cases. 7 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.