What problem does it solve?
This Skill automates the assembly of a complete LaTeX manuscript by drafting missing paper sections (Abstract, Methodology, Experiments, Conclusion), converting raw experimental logs into publication-ready tables, and integrating figures and citations while preserving existing template content.
Core Features & Use Cases
- Single-call multimodal drafting: Compose one comprehensive multimodal LLM prompt that produces the filled LaTeX manuscript to preserve global coherence across sections.
- Deterministic extraction & gating: Run scripts to parse numeric tables from experimental_log.md, validate LaTeX structure, and verify citation keys to prevent hallucinated numbers or orphan citations.
- Figure and citation integration: Inspect or consume provided PNG figures and captions, include them with exact filenames, and use the citation_pool and refs.bib to ground assertions about related work.
- Use Case: Finish an ML conference paper by merging Intro+Related Work with newly generated Methodology, Experiments, and Conclusion, producing workspace/drafts/paper.tex that compiles without float-placement or citation errors.
Quick Start
Ask the agent to "Write the methodology and experiments sections using outline.json, experimental_log.md, figures/, and citation_pool.json, then save the compiled LaTeX to workspace/drafts/paper.tex".