01 / Literature
Build a library you can check
Add your PDFs and notes, then look for related work. Sort papers by how they help: foundational work, close alternatives, and baselines. Use the library to keep track of full texts and the papers you have marked for close reading.
Read the original layout, highlight or annotate a passage, or ask about a specific paragraph. The research reader can translate in place. For a continuous reading copy, open the translation in LANCER READING.
02 / Questions
Look for the place where the explanations disagree
Put papers side by side: what they assume, how they test, and where their methods fail. The challenge analysis points to the relevant passages. Choose a question card and make it the thread for the next stages; add more literature if the first comparison leaves a gap.
03 / Ideas
Try more than one way to answer it
The idea draw produces alternatives for the same question. Athena goes deeper: it compares candidates against prior work, checks quoted evidence, and asks what would distinguish one explanation from another. Its judgments help you choose what to investigate next.
The Fusion Engine lets you bring several idea cards together. Inspect where they complement each other, then review proposed combinations against their starting ideas and a simple combination.
04 / Derivation
Write down what has to be true
Lay out the definitions and assumptions, then map which statements depend on which. Review the derivation for missing steps and check that cited evidence actually supports it. Leave unproved claims marked as conjectures, with a test that could distinguish them.
When a concept gets in the way, ask the research tutor about the current paper, card, or stage. It can explain how the idea is used here, show a derivation, and check your understanding with a short question.
05 / Experiments
Compare a change against a baseline
Compare a small, standard, or more ambitious design before running one. Start with the baseline, change one part at a time, and record each run. Keep changes that help, revert those that do not, and retain the evidence for both.
Pause, redirect, or stop a stage while it runs. The project keeps its progress and files so you can inspect the result and continue at the point you left.
06 / Audit & writing
Write only what the evidence can support
The audit compares the theory and experimental design with what was actually run and observed. Mark the claims that are supported and the ones that still need work. Then take the outline into a draft you can edit in Markdown or LaTeX.
In step-by-step mode, inspect each stage before it feeds the next one. You can revise, rerun, or skip it. The development build also has a managed mode; you can switch back to reviewing stages yourself. Neither mode decides whether a research claim is true for you.
Research team
Split the research. Keep a shared record.
Use a pairing code to join the same research project and assign stages to different research stations. A stage runs on the computer of the member responsible for it. The team can see who is working on which stage, how experiments are progressing, and which server is in use.
The development build supports sharing through a shared folder or a direct connection. When both sides edit the same material, it tries to merge the changes and keeps a separate copy of unresolved conflicts so the team can review them.
Starforge & onboarding
Choose how you want to see the research
The standard view lays out the research stages, literature, cards, and tutor. Starforge presents the same project in a game-like chapter map, with milestones based on the work done in the project. Both views have an illustrated walkthrough you can reopen from the light bulb in the toolbar.
What you can keep
Papers and annotated notes, question and idea cards, derivations, experiment records, an audit report, and an editable draft.
