Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →brycewang-stanford/asplos-writing-style
Use when drafting or revising ASPLOS prose — engineering the first two pages that rapid review actually reads, stating the cross-layer insight as one quotable sentence, structuring per-layer mechanism sections, writing bounded claims, and fitting everything into 11 self-contained pages that include figures and footnotes.
Passed 20 of 22 test cases with the skill, and 13 of 22 without it.
and 4 more in the full report
7 fixed13 passed either way2 still failing0 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 7 test cases the skill fixed. Same model, same prompt, with and without the skill.
What is the requirement for author name formatting in ASPLOS paper references?
| Checked | Without skill | With skill |
|---|---|---|
| States that every author's full first and last name must be listed | Missed | Passed |
The instructions your model reads when the skill loads.
Two 2027 mechanics dictate ASPLOS prose (CFP, checked 2026-07-08): the rapid review reads only the first two pages, and the 11-page limit counts all text, figures, tables, and footnotes — references excluded. The first makes pages 1-2 a survival organ; the second makes every figure a budget decision. Style here is mostly resource allocation.
By the bottom of page 2, a screener must hold four things. Audit for them mechanically, in order:
| Element | Test | Common failure |
|---|---|---|
| Cross-layer problem | Can a reader from any of the three communities restate it? | A single-community problem with "co-design" sprinkled on |
| One-sentence insight | Is there a sentence a screener could quote in their note? | Insight diffused across three paragraphs, or a category ("we co-design X and Y") instead of an idea |
| Mechanism sketch | Does the reader know what each layer does? | Architecture described, software hand-waved (or vice versa) |
| Headline evidence | Result + platform class (silicon/FPGA/simulator) + baseline named? | "Significant improvements on a variety of workloads" |
Corollary: background sections, notation, and roadmaps do not belong on pages 1-2.
The worked example (../../resources/worked-examples/01-introduction.md) shows a
full before → after against exactly this table.
The load-bearing sentence in an ASPLOS paper names information or capability that crosses a layer boundary: what the hardware knows that the OS needs, what the compiler can prove that the microarchitecture can exploit. Drafting discipline:
Weak: "We propose a hardware/software co-designed approach to memory tiering."
(names a category and a topic; contains no idea)
Strong: "The contention data the OS needs already passes through the memory
controller on every access; the fix is a narrow architectural channel,
not a smarter kernel heuristic."
(names the information, its current location, and the boundary moved)
If this sentence cannot be written, the problem is usually fit, not prose — return
test cases passed, vs 13 / 22 without
Published skills whose descriptions closely match this one, with their headline benchmark lift.
More skills from the same source and nearby in the registry, with their headline benchmark lift.