claude-all-in-one-plugin

All-in-one Claude Code plugin: end-to-end pipelines for data analysis (profile → clean → EDA → validate → report), leakage-safe tabular ML modeling (frame → split → baseline → select → tune → evaluate → drift/readiness), paper-reading research (discover → digest → link), Kaggle agent-competition building (frame → scaffold → baseline → self-play → profile), authorized agent-security red-teaming (frame → scaffold → chains → eval → harden), object-detection computer vision (inspect → split → scaffold-train → evaluate → submission), and rigorous AI-agent research production (frame → design → validate-task → scaffold-trials → analyze → diagnose → findings) — plus a reactive code-quality layer, driven by conductor agents that stop at human-decision checkpoints and stay honest about what offline tooling can't measure (they scaffold and evaluate; they don't train models on GPU-hours in-session).