Use when planning an ATC (ACM SIGOPS Annual Technical Conference, formerly USENIX ATC) campaign end to end — running the year backward from the early-June deadline through the two review rounds, the rebuttal, conditional-acceptance shepherding, the post-acceptance artifact evaluation, and the ACM Open Access camera-ready for the November Hong Kong conference.
SkillScore78Maintained
Automatic
◆Research pack
Last verified 8/15/2026
What it changes
Passed 18 of 23 test cases with the skill, and 6 of 23 without it.
Without skill26%
With skill78%
Checks it passed that the model alone missed
Advises freezing system experiments.
Notes that ACM SIGOPS rules govern ATC 2026.
Recommends creating a claim-to-experiment map.
and 16 more in the full report
13 fixed5 passed either way4 still failing1 broke
Pass-rate lift
+52 pp
Tokens
+47%
Agent turns
0%
Response time
-28%
Model gemini-3.6-flashMeasured 8/15/2026v118 of 23 cases passed with skill
Tokens, turns and time describe resource use, not a dollar price.
See it in action
One of 13 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 13
Prompt
What is the targeted submission month and timezone for ATC 2026 initial paper upload?
What was checked, without and with the skill
Checked
Without skill
With skill
Identifies early June 2026 as the submission timeframe.
Missed
Passed
Identifies Anywhere on Earth (AoE) as the deadline timezone.
Passed
Passed
case-18 · gemini-3.6-flash · 8/15/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md85 linessingle file
ATC Workflow
Plan the campaign backward from the deadline. ATC 2026 — the first ACM SIGOPS edition of the
former USENIX ATC — runs on a calendar and a review structure that both changed in the
transition: an early-June 2026 submission deadline (AoE; 1 vs. 10 June unresolved across
renderings — 待核实), a two-round, extended-abstract review, conditional acceptance with
shepherding, a separate artifact-evaluation track, and the conference in Hong Kong, 15-18
November 2026. Reopen the SIGOPS call for the exact dates before committing to any of the offsets
below.
The arc (backward from the deadline)
[T-0 Submission] paired upload: 2-page extended abstract + full paper on atc26.hotcrp.com
[Round one] two reviewers judge each extended abstract; below-bar papers cut early
[Round two] 3-4 reviewers read the full paper of survivors
[Rebuttal] ~1 page: correct factual misreadings; note prior-submission changes if any
[Decision] conditional accept / reject
[Shepherding] a PC shepherd signs off required revisions before final acceptance
[Artifact evaluation] separate post-acceptance AEC track -> Available / Functional / Reproduced
[Camera-ready] de-anonymize, satisfy shepherd, ACM Open Access metadata + production checks
[Nov 15-18] present at the Hong Kong conference; talk video/slides per the venue norm
Backward plan from the June deadline
T-10 to T-8 weeks: lock the system and freeze the experiments. On a systems paper the testbed
and measurements are the long pole; leave prose churn for later. Pin environments now
(atc-reproducibility) — provenance cannot be reconstructed at the deadline.
T-6 weeks: full-paper draft complete against atc-writing-style; baselines and honest costs
in place per atc-experiments.
T-3 weeks: write and stress-test the two-page extended abstract — the round-one decision
surface. Have someone read only the abstract and tell you the problem, contribution, testbed,
and result. If they cannot, fix it.
T-2 weeks: build the anonymized, runnable artifact and the claim-to-experiment map.
T-1 week: double-blind sweep on the final PDF, abstract, and artifact; fill HotCRP fields and
conflicts; confirm the exact AoE deadline day.
T-0: upload both the extended abstract and the paper; re-download and read cold.