Appendix D. Further reading¶
Learn by doing (tutorials)¶
- First figure (15 min): make a two-panel figure from example data with commands alone, and fix a check warning in the config.
- A microscopy figure: make a figure of representative images + quantification from public microscope images, with the image adjustment record and a statistics table.
Do a task (how-to)¶
- Connect AI agents: connect the fig-atlas tools to Claude Code, Gemini CLI, opencode, Antigravity, Cursor or Codex.
- Write a domain pack: add the data formats and figures of another field, such as electrophysiology or imaging (for developers).
Look up (reference)¶
- Data bundle contracts: the columns and format a bundle needs for each kind of figure.
- Renderer inventory: the kinds of figure, and which projects use them.
- Command reference: every command and option (generated).
- Python reference: the functions analysis scripts and figure specs call, by task (generated).
- Config reference: the config keys of each kind of figure, with defaults, units and explanations (generated).
Why it is built this way (explanation)¶
- Design decisions: drawing at printed size, the config as the source of truth, figure shapes separate from fields, commands and AI, one place for the rules.
- The full rules (
AGENTS.md): every rule people and AI follow.
Previous: Appendix C. Commands on one page · Back to the start: 0. How to read this guide