Standardized M2G-NET experiment pipeline: data check → train → evaluate → save results to docs/. Use when running training, ablation, baseline comparison, gate stability, or any multi-phase experiment on M2G-NET.
Spawn an isolated math-checker subagent to independently verify that src/ code matches docs/MATHEMATICAL_FORMULAS.md. Use after any edit to fusion.py, model.py, interaction.py, or loss.py.
Validate that M2G-NET code (src/fusion.py, src/model.py, src/interaction.py) matches the documented math in docs/MATHEMATICAL_FORMULAS.md. Use whenever editing model architecture, loss, gate computation, calibration, or metrics.
Spawn an isolated result-analyst subagent to read docs/ and outputs/ and produce a structured experiment summary with metric tables, key findings, and next-step recommendations.
Rate-limit-resilient pipeline with checkpoint/resume for long multi-phase sessions. Saves progress to .claude/pipeline-state.json after each phase. Use when starting a complex multi-phase task that risks hitting rate limits, when resuming an interrupted…
Act as the world's most rigorous scientific editor for transportation and green transition research. Use this skill when the user asks for critical review, peer review feedback, manuscript critique, methodology assessment, or wants to improve scientific…