Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when preparing the camera-ready of an accepted ACM CoNEXT paper for its assigned PACMNET issue — systematically de-anonymizing, completing PACMNET journal metadata and ACM rights/CCS/keywords, permanentizing artifact and DOI links, applying any shepherd-required changes, and passing ACM production checks.
.claude/skills/brycewang-stanford-conext-camera-ready/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -19% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 27% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -17% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 36% | 0% |
| case-14 | ✗→✓ | ▲ Improved | -21% | 0% |
Deliver the camera-ready for the PACMNET issue your paper was scheduled into. This is the step where the double-anonymous manuscript becomes a citable ACM journal article, so two things dominate: systematic de-anonymization and complete journal metadata. If your paper went through a one-shot revision or a shepherd, this is also where you confirm every required change actually landed before the deadline (December cycle: camera-ready 30 Apr 2026; major-revision path 31 Jul 2026 — verify your cycle's date).
The review copy was scrubbed of identity; the camera-ready must restore it completely and only where appropriate:
text[Authors + affiliations] add the real author block and affiliations [Acknowledgments] restore funding, grants, and collaborators [Self-citation] switch "prior work [12]" back to first person only where accurate [System/testbed names] restore real names of systems, testbeds, deployments [Artifact links] swap anonymized URLs for the permanent DOI-issuing archive [Repository] point to the real, public repository (not the anon service)
Do a dedicated pass: a leftover "Anonymous" or a dead anon-service link in the published article is a visible production error.
Because CoNEXT publishes in PACMNET, the camera-ready carries journal metadata, not just conference proceedings fields:
confirm the exact license and copyright string from your rights form.
article-number/DOI fields the production kit requests.
Software Heritage) and cite that DOI in the paper.
are correctly associated (coordinate with the reproducibility committee; see conext-artifact-evaluation).
locations.
minimum necessary changes against the final PDF; the shepherd signs off before publication.
edits to the required changes, de-anonymization, and polish.
acmart version/options for PACMNET; do not alter margins or fonts.production/TAPS pipeline enforces.
review limit — verify).
text[De-anon] authors, acks, system/testbed names, artifact links all restored? yes/no [No leftovers] no "Anonymous", no dead anon-service links? yes/no [Metadata] CCS + keywords + rights block + ORCIDs + PACMNET issue/DOI complete? yes/no [Availability] artifact on a DOI archive; badge association correct? yes/no [Shepherd] required changes confirmed in the final PDF? yes/no/n-a [Production] acmart version, ACM PDF checks, page allowance, figures? pass/fail
text[Camera-ready status] ready / blocked [Cycle + issue] <December | June> cycle -> PACMNET <volume/issue> [De-anonymization] complete / leftovers: <where> [Metadata] CCS/keywords/rights/ORCID/DOI status [Availability] permanent artifact DOI + badge association [Production] acmart + ACM checks pass? outstanding items with the deadline
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 22,939 | 13,168 | -43% | 1 | 1 | 0% | 2,930 | 2,385 | -19% | 0 | 0 | — |
case-02 | fail→pass | 17,306 | 16,579 | -4% | 1 | 1 | 0% | 2,313 | 2,947 | +27% | 0 | 0 | — |
case-03 | fail→pass | 29,058 | 18,961 | -35% | 1 | 1 | 0% | 3,982 | 3,323 | -17% | 0 | 0 | — |
case-04 | pass→fail | 28,499 | 21,924 | -23% | 1 | 1 | 0% | 3,819 | 3,966 | +4% | 0 | 0 | — |
case-05 | pass→fail | 29,454 | 20,085 | -32% | 1 | 1 | 0% | 3,201 | 3,448 | +8% | 0 | 0 | — |
case-06 | pass→pass | 19,243 | 13,204 | -31% | 1 | 1 | 0% | 2,273 | 3,008 | +32% | 0 | 0 | — |
case-07 | fail→pass | 17,432 | 10,220 | -41% | 1 | 1 | 0% | 1,976 | 2,682 | +36% | 0 | 0 | — |
case-08 | pass→pass | 14,824 | 12,133 | -18% | 1 | 1 | 0% | 1,578 | 2,325 | +47% | 0 | 0 | — |
case-09 | pass→pass | 11,614 | 11,689 | +1% | 1 | 1 | 0% | 2,001 | 2,287 | +14% | 0 | 0 | — |
case-10 | pass→pass | 17,560 | 12,751 | -27% | 1 | 1 | 0% | 1,741 | 2,265 | +30% | 0 | 0 | — |
case-11 | pass→pass | 15,481 | 26,339 | +70% | 1 | 1 | 0% | 1,639 | 2,563 | +56% | 0 | 0 | — |
case-12 | pass→pass | 30,285 | 14,499 | -52% | 1 | 1 | 0% | 2,197 | 2,499 | +14% | 0 | 0 | — |
case-13 | pass→pass | 19,337 | 15,551 | -20% | 1 | 1 | 0% | 2,038 | 2,536 | +24% | 0 | 0 | — |
case-14 | fail→pass | 15,919 | 9,450 | -41% | 1 | 1 | 0% | 2,223 | 1,766 | -21% | 0 | 0 | — |
case-15 | fail→pass | 23,456 | 7,426 | -68% | 1 | 1 | 0% | 3,272 | 1,385 | -58% | 0 | 0 | — |
case-16 | pass→pass | 17,472 | 13,016 | -26% | 1 | 1 | 0% | 1,677 | 2,240 | +34% | 0 | 0 | — |
case-17 | pass→pass | 16,730 | 9,239 | -45% | 1 | 1 | 0% | 1,804 | 2,144 | +19% | 0 | 0 | — |
case-18 | pass→pass | 18,330 | 21,154 | +15% | 1 | 1 | 0% | 2,913 | 3,289 | +13% | 0 | 0 | — |
case-19 | pass→pass | 18,121 | 16,535 | -9% | 1 | 1 | 0% | 2,391 | 2,656 | +11% | 0 | 0 | — |
case-20 | pass→pass | 18,124 | 13,948 | -23% | 1 | 1 | 0% | 1,697 | 2,571 | +52% | 0 | 0 | — |
case-21 | pass→pass | 19,557 | 8,012 | -59% | 1 | 1 | 0% | 1,817 | 2,231 | +23% | 0 | 0 | — |
case-22 | fail→pass | 9,244 | 12,293 | +33% | 1 | 1 | 0% | 1,173 | 1,913 | +63% | 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. 22 cases were attempted. The headline lift of +23 percentage points is the difference between those two pass rates over the 22 comparable cases. 2 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.