The wall - why the model cannot open your instrument file
Lesson 1 of 6 · What the model can and cannot see
1 · Learn the move · RTO. Role · Task · Output
Week one goes like this: you attach the .fcs from the cytometer or the .raw from the mass spec, nothing useful happens, and you conclude the tool is a toy. The wall is real - vendor binaries are not readable by any model - but the conclusion is wrong, because the model was never going to be the reader of your file. It is the author of the path to reading it. Two corollaries before anything else. A screenshot of the trace is not access: an image of a gel, blot, or chromatogram never produces a number, only a description. And role-task-output is how you ask for the path: role, a conversion planner; task, name the format and instrument; output, whether anything reads it directly, what converts it, and what to hand back afterward.
You are a scientific data format planner. My file: [extension] produced by [instrument, software, version if known]. Output exactly three sections: (1) DIRECT - can any model read this file as-is: yes, structure-only, or no; (2) PATH - the standard open-source route to an open format, named tools only, no invented ones; (3) HAND BACK - what small text summary of the converted file I should return for reasoning. If you are not certain a tool handles this exact variant, say so - vendor formats have versions.
2 · Your turn. You write the prompt
A new rotation student spent an hour attaching a flow cytometer .fcs file to a chat window in different ways, got three cheerful non-answers, and told you the whole thing is useless - which is what you suspected too. Write the prompt you should both have started with: one that asks for the path, not the file.
Remember: the AI sees only your prompt, not this page. If the situation isn't in your prompt, it doesn't exist.
Optional. These shape the output when you run your prompt below, not your score.