agentsclimarketplace

Issta supplementary

Skill brycewang-stanford/Awesome-Journal-Skills/ISSTA-Skills/skills/issta-supplementary

Use when deciding what lives in the ISSTA 18-page body versus the artifact and any appendix, covering the no-unlimited-appendix reality, what reviewers will and will not open, splitting a testing/analysis paper between body and package, anonymity of supplementary material, and keeping decision-critical evidence inside the reviewed pages.From its SKILL.md

Install
npx -y skills add brycewang-stanford/Awesome-Journal-Skills --skill issta-supplementary

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

SKILL.md

4.0 KB, 797 tokens by cl100k_base, as published. Nobody here has run it

ISSTA Supplementary

Use this when deciding where each piece of content belongs. ISSTA's 18-page limit counts appendices against the total, so there is no unlimited appendix to absorb overflow: the body must stand on its own, and the artifact — not a separate PDF supplement — is where extra material goes. Reopen the current call to confirm whether any in-PDF appendix or separate upload is permitted this cycle.

The split rule

  • Everything decision-critical lives inside the 18 pages: the threat model, the technique, the evaluation design, the headline results, and the threats to validity. If a reviewer needs it to judge the paper, it cannot live only in the artifact.
  • The artifact carries what supports but does not replace the argument: the full tool, complete per-subject tables, extra configurations, raw logs, and regeneration scripts.
  • Do not hide core method steps or primary results in an appendix or the artifact to make the body fit. Reviewers grade what is in the reviewed pages; buried evidence is invisible evidence.
  • Because appendix pages count against 18, a long appendix competes directly with the body. Prefer moving detail to the artifact over spending page budget on it.

What reviewers open, and what they do not

ContentWhere it belongsReviewer attention
Technique definition, threat modelBodyRead closely — graded for soundness
Headline results + statisticsBodyRead closely — graded for evaluation
Threats to validityBodyExpected; its absence is noticed
Full per-subject result tablesArtifactSkimmed if a body claim is contested
Raw fuzzing/analysis logsArtifactRarely opened; ship for completeness
The runnable toolArtifact (evaluated separately)Run by artifact evaluators, not always by PC

Assume a PC reviewer reads the 18 pages and dips into the artifact only to resolve a doubt. Design the body so no such dip is required to accept the paper.

Anonymity of supplementary material

  • Keep the review-time artifact double-anonymous: no repository owner, commit authors, container labels, institution paths, or acknowledgements.
  • Verify archives open on a clean machine with no credentials, caches, or OS metadata baked in.
  • An anonymized public mirror or an anonymous upload is fine; a link to your named GitHub is a double-anonymous violation.

Vignette: splitting a static-analysis paper

A submission presenting a new taint-analysis with a soundness argument and a large evaluation: the body keeps the analysis rules, the soundness sketch, the evaluation protocol, and the two decision-critical result tables (precision/recall against ground truth, and a comparison to one established baseline); the artifact holds the full implementation, the complete per-benchmark tables, alternative sensitivity configurations, and the scripts that regenerate every table. Nothing needed to accept the paper lives only in the artifact, because artifact inspection by the PC is discretionary.

Body (<= 18 pp)          Artifact
- analysis rules          - full analyzer implementation
- soundness sketch        - complete per-benchmark tables
- evaluation protocol     - extra sensitivity configs
- 2 headline tables       - table-regeneration scripts + raw logs
- threats to validity     - pinned subjects (SHAs)

Output format

[Split status] Ready / Needs fixes / Not ready
[Body carries] <decision-critical items present in the 18 pages>
[Artifact carries] <supporting material>
[Page-budget risk] <appendix competing with the body?>
[Anonymity checks] passed / issues
[Main-paper dependency] <what breaks if the artifact is never opened>

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.

Keep looking

Skills are one crate of 325,949. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.