Workflow-first research skill pack for computer systems, networking, and AI papers. Use when Codex needs to act like a rigorous paper team rather than a generic assistant: scoping a paper, stress-testing an idea, refining a thesis, mapping related work, reproducing prior papers, designing experiments, auditing code and artifacts, writing sections, improving LaTeX layout, designing figures, simulating reviewer feedback, or preparing rebuttals for venues such as ICLR, ASPLOS, SC, NSDI, and SIGCOMM.
Reproducibility and evidence-audit skill for research projects. Use when Codex should trace figures and claims back to scripts, configs, datasets, seeds, hardware assumptions, and logs; find missing provenance; or prepare the code and experiment pipeline for submission, open-sourcing, or artifact evaluation.
Research-code review skill for systems, networking, and AI projects. Use when Codex should review experiment code, simulators, training pipelines, benchmarks, data processing, or plotting logic with emphasis on correctness, reproducibility, and whether the implementation really supports the paper's claims.
Experiment-planning skill for research papers in systems, networking, and AI. Use when Codex must design or audit baselines, metrics, workloads, ablations, statistical checks, scaling studies, sensitivity analysis, and failure tests so that a paper's claims are actually supported.
Experimental-process management skill for research projects. Use when Codex should structure runbooks, logging, checkpointing, seed control, result triage, failure analysis, experiment queues, and evidence collection so that the day-to-day experimental workflow stays publication-grade instead of devolving into ad hoc trial-and-error.
Figure and table design skill for research papers. Use when Codex should decide what to visualize, sketch figure plans, critique existing charts, improve captions, structure result tables, or make visual evidence easier for strict reviewers to understand in AI, systems, or networking papers.
Topic-finding skill for research projects in computer systems, networking, and AI. Use when the user has a broad research area, venue target, resource constraint, or personal interest but does not yet have a one-paper thesis. This skill must scan recent papers, generate candidate topics, run each candidate through researchstack-idea-review, and only surface the surviving paper-sized ideas.
Guided idea-refinement skill for research projects that are interesting but not yet submission-ready. Use when Codex should improve a paper idea through structured questioning, narrowing, hypothesis building, contribution shaping, and experimental reframing instead of only approving or rejecting it.