← ClaudeAtlas

analytical-figureslisted

Publication-grade figures and the gated analysis behind them for analytical chemistry / spectroscopy work. Use for ANY of: plotting FTIR / IR / ATR or PXRD / XRD spectra (overlays, waterfalls, baseline correction, band integration, %RSD/precision); building calibration curves (linear regression, confidence/prediction bands, residuals, LOD/LOQ, recovery, ICH Q2); generic line/bar charts for lab data; or turning raw instrument files (.spc, xy/csv) into journal-ready PDF+PNG. Trigger on: spectrum, spectra, FTIR, IR, ATR, PXRD, XRD, diffractogram, waterfall, baseline, band area, integration, RSD, calibration curve, linearity, LOD, LOQ, residuals, recovery, publication figure, journal figure; crystal structure, CIF, .cif, space group, hydrogen bond, displacement ellipsoid / ORTEP, packing diagram, calculated PXRD from a CIF, structure validation table.
TheDop/analytical-figures-skill · ★ 1 · Data & Documents · score 75
Install: claude install-skill TheDop/analytical-figures-skill
# analytical-figures A figure + analysis skill for spectroscopy and analytical QC. It assumes you already know matplotlib. Its job is the two things that actually go wrong: **the numbers are subtly wrong before they ever reach a plot**, and **the figure looks "fine" but is broken in a way a quick glance misses**. So it forces verification and it locks a house style. Nothing here is a plotting tutorial. If a request is just "how do I do X in matplotlib", answer it directly; this skill is for producing *deliverables* from analytical data. ## Output contract: a self-contained script, not a figure The deliverable this skill produces is **one runnable, self-contained Python script** — a CONFIG block plus the analysis body — that the user runs to generate the figure. Do **not** render the figure in-session and hand over a PNG; hand over the *script that produces it*, so it is reproducible, editable, and attachable to a report. The modules in `scripts/` are the maintained source of truth; you write the analysis against them, then `bundle.py` amalgamates the modules it uses + your body into a single file that depends only on the scientific stack (numpy, matplotlib, optionally scipy/scienceplots), not on this skill being installed: ``` python bundle.py my_analysis.py -o standalone.py # hand over standalone.py ``` **Provenance is stamped, not narrated.** `bundle.py` writes a provenance header into the deliverable — skill version, the Python + scientific-stack versions it was b