Analyze ICLR reference paper LaTeX structure
Takeaway: I'd tell it upfront that the main .tex file might be large and to just use offset/limit reads from the start instead of hitting the token-limit error first.
Stack · 7 sessions
latex sessions with coding agents
7 sessions on latex with a coding agent: the agent failed at least once in 6, and 1 finished without a human stepping in.
Analyze old Euro-Par HyperCluster LaTeX paper structure
Takeaway: I'd start the same way next time since it's a simple read-and-summarize task with no code changes needed.
Regenerate Fig 2 with PaperBanana and polish LaTeX paper
Takeaway: I'd start by narrowing the ask to just Fig 2 and skip asking it to explore the whole paper and tool in one go, since that's what blew the context twice.
Regenerate Figure 2 ring pipeline diagram with PaperBanana
Takeaway: Next time I'd tell it upfront to keep the image context small since loading multiple full-resolution figure iterations blew past the model's token limit right at the finish line.
Regenerate Figure 2 and polish LaTeX paper with PaperBanana
Takeaway: Next time I'd start by manually trimming what context gets loaded, since the agent kept blowing past the 128k token limit by dragging in the whole paper, config, and image files at once.
Add figures, Llama-3.2-3B data, and recompile Euro-Par LaTeX paper
Takeaway: Next time I'd tell it upfront which python environment has matplotlib and paperbanana installed, since it burned a lot of turns discovering that the system python, the project venv, and pip were all broken before it found the paperbanana venv that actually worked.
Rewrite HyperCluster Euro-Par paper and generate figures with PaperBanana
Takeaway: Next time I'd keep the session shorter and more scoped — splitting 'update the paper text' from 'generate all figures and recompile' into separate sessions so the context window doesn't balloon and kill the final compile step.