--- name: brycewang-stanford/cav-workflow source: https://app.decimal.ai/s/brycewang-stanford-cav-workflow@1/SKILL.md source_sha256: 44537be01f1d --- # CAV Workflow Use this as the project-management skill for a CAV submission. Replace every date with the current official timetable and work backward from the AoE paper deadline. CAV publishes open access in **Springer LNCS**, reviews in **two stages** (an early-reject filter, then a rebuttal), and runs an **optional** artifact evaluation after notification — plan for all three, not just accept/reject. CAV is a conference with **no standing editor-in-chief**; the rotating leadership is the per-edition Program Chairs and Steering Committee. For CAV 2026 the Program Chairs are **Anthony W. Lin, Eva Darulova, and Philipp Rümmer** (verified 2026-07-09), and the venue is **Lisbon, part of FLoC 2026**. Chairs rotate yearly; re-check the organization page rather than carrying a name forward. ## Milestones - **Venue + category fit:** confirm CAV over TACAS/FMCAD/VMCAI and pick the category (Regular / Tool / Application / Industrial) (`cav-topic-selection`). - **Evidence lock:** freeze the theorem/algorithm, the benchmark set (revision pinned), the baseline tools and versions, the resource limits, and the artifact contents. - **Submission:** upload the LNCS-formatted PDF in the right category, anonymized if Regular or Application, with the **artifact-intent declaration** set (CAV 2026: 28 Jan 2026 AoE). - **Stage-1 filter:** each paper gets two reviews; weak papers are rejected early (tentative first-round outcome ~4 Mar 2026). - **Rebuttal:** papers passing stage 1 get two more reviews and an author-response window (CAV 2026: 30 Mar - 2 Apr 2026). - **Notification:** accept/reject (CAV 2026: 17 Apr 2026); archive all reviews and versions. - **Artifact evaluation (optional):** register and submit to the AEC on its own deadline (CAV 2026 artifact registration: 22 Apr 2026), pursuing Available / Functional / Reusable badges. - **Camera-ready:** prepare the LNCS open-access final version and complete Springer metadata and rights; present at the conference (CAV 2026: 26-29 Jul 2026, Lisbon / FLoC). ## Backward plan from the submission deadline | Weeks out (heuristic) | Verification-paper milestone | |---|---| | 10+ | Technique/algorithm fixed; the soundness/completeness statement drafted | | 8 | Tool implemented and stable; benchmark set and baseline tools/versions chosen and pinned | | 6 | Full evaluation run under fixed resource limits; per-instance data and logs archived | | 4 | Full draft in `llncs`; proofs complete (in body or clearly-marked appendix); artifact assembled | | 3 | Internal mock review by a verification-literate reader; differential/soundness cross-check done | | 2 | Scope and limits sharpened; related-work delta tightened; page limit met | | 1 | Anonymity sweep on PDF and artifact (Regular/Application); artifact-intent decided | | 0 | Upload in the correct category via the live portal; re-download to confirm | These offsets are planning heuristics only — anchor every one to the current Important Dates page, never to a previous cycle's calendar. ## The two-stage cycle: plan for the early reject CAV's stage-1 filter means a paper can be **rejected before the rebuttal ever happens**. Two planning consequences: - **The first two reviews must be able to defend the paper on their own.** A contribution that only becomes convincing after the rebuttal may never reach stage 2. Make the soundness and the headline benchmark result legible in the first read. - **Missing the annual deadline costs roughly a year at CAV.** A sibling flagship (TACAS in spring, FMCAD, VMCAI) may have a nearer date — factor that into routing rather than idling (`cav-topic-selection`). ## Failure modes by stage - **Tool still unstable at week 6** forces a rushed evaluation nobody audited — the classic reproducibility/soundness doubt in the making. - **Benchmarks or baselines unpinned** means the comparison cannot be reproduced and the reviewer discounts the numbers. - **Leaving anonymization to the final week** (Regular/Application) is how a solver name or a benchmark path leaks identity under double-blind review. - **Treating stage 1 as a formality** and saving the strongest framing for the rebuttal forfeits the paper before the rebuttal exists. - **Assuming artifact evaluation is required** — it is optional and non-conditional at CAV; plan it, but do not block the camera-ready on it. ## Coordination notes - Assign one owner for the artifact (benchmarks, versions, seeds, resource limits) and another for the anonymity sweep; shared ownership is how both slip. - Archive the exact submitted PDF, the artifact, and (later) the rebuttal — a stage-2 review and any camera-ready must quote them precisely. ## Output format ```text [Current stage] idea / evidence / writing / submitted / stage-1 / rebuttal / accepted / artifact [Next official deadline] [Critical path] [Risk register] [Owner map] person or role> ```