wissingcc

You can search and capture literature with citation-graph expansion, validate whether a research idea is worth pursuing, run and checkpoint experiments with convergence accounting, interpret results with statistical verdicts, write CS/AI/robotics manuscripts, and produce HTML slide decks and reports.

They target academic researchers, PhD students, and ML/robotics engineers who manage multi-stage research projects — from literature review and idea convergence through experiment execution, evidence interpretation, and final paper or presentation delivery.

research_manager anchors the .kilo/ project skeleton and file lifecycle; all other skills declare it as a dependency. knowledge_keeper, research_progress, experiment_manager, result_analysis, and result_visualization each own one research phase and explicitly delegate to one another.

Every skill except research_manager and plain_talk declares 'requires: research_manager', so the project skeleton skill must be present first. plain_talk must load at the start of every session to enforce jargon-free, conclusion-first output.

The skill_rsi skill acts as a self-improvement inbox: it records friction points, user corrections, and missing scenarios as proposals, applying changes only after explicit user review — it never silently edits other skills.