What problem does it solve?
Writing a well-structured long-form article requires an outline grounded in actual research rather than guesswork. This Skill drafts an outline from parametric knowledge and then refines it against real research conversations, so the final structure reflects what was actually learned.
Core Features & Use Cases
- Two-pass outline generation: Drafts an outline from the model's prior knowledge, then refines it using the research conversation history to merge, add, or drop sections.
- Pipeline integration: Runs as phase 2 of the STORM pipeline, consuming
research/conversations.jsonl and research/sources.json and producing outline.md.
- Idempotent execution: Skips regeneration when
outline.md already exists unless --force is passed, and records phase status in run-config.json.
- Use Case: After completing STORM research on a topic like "quantum error correction", invoke the outline phase to produce a refined multi-section article skeleton ready for the writing phase.
Quick Start
Ask the assistant to generate an outline for your STORM article on a given topic after the research phase has completed.