CROS is an AI-native skill system for complex systems research, disaster science, earthquake engineering, urban resilience, and scientific writing.
This repository is designed for Codex Skills using the open SKILL.md convention. Skills are stored under .agents/skills/ so they can be used as repository-scoped skills. You can also install them globally to ~/.agents/skills/.
CROS is not a prompt collection. It is a reusable research workflow layer for:
- scientific reasoning,
- literature and evidence synthesis,
- mechanism-oriented model building,
- reproducible scientific coding,
- academic paper architecture,
- complex systems and disaster-domain reasoning,
- project memory.
From this repository:
bash scripts/install_to_user_skills.shOr manually:
mkdir -p ~/.agents/skills
cp -r .agents/skills/* ~/.agents/skills/Restart Codex if the skills do not appear immediately.
$scientific-thinking Help me convert this observation into testable hypotheses.
$complex-systems-research Analyze this urban disaster problem through state, coupling, feedback, critical transition, and intervention.
$paper-architecture Build a Nature-style paper storyline from these results.
$earthquake-wavefield-research Review this waveform-network analysis pipeline.
$reviewer-response Draft a concise response to this reviewer comment.
core-scientific-thinking: observation → inference → mechanism → prediction → intervention.evidence-check: claim classification, evidential strength, uncertainty, falsification.hypothesis-design: pattern → alternative mechanisms → testable hypotheses.literature-review: research-map style synthesis, not annotated bibliography.paper-architecture: high-level manuscript story structure.nature-writing: restrained, evidence-forward scientific prose.reviewer-response: acknowledge, clarify, revise, evidence.python-scientific-computing: reproducible Python research coding.figure-design: publication-grade scientific figures.complex-systems-research: state, coupling, feedback, emergence, criticality, control.critical-transition: CSD, flickering, dimensional collapse, tipping, validation.network-science: dynamic, multiplex, interdependent, temporal network reasoning.earthquake-wavefield-research: dynamic waveform correlation networks and spectral indicators.openquake-psra: PSHA/PSRA workflow discipline.urban-resilience-m3: M3 / urban system state, cascade, intervention, recovery.project-memory: project-level research state tracking.
This is a v1.0 starter skeleton. It is intentionally modular: each skill is focused on one job and can be edited independently.