scienceplot-py
Skill4 starsChanged 20 days agoMITView on GitHub
Generate Python matplotlib plotting scripts in the user's scienceplots lab style: with plt.style.context(["science", "nature"]), pparam = dict(...), raw-string LaTeX labels, and fig.savefig(..., dpi=300, bbox_inches='tight'). Support parquet, CSV, NPY/NPZ data and single-line, multi-line, scatter/errorbar, or subplot variants. Generate both the reproducible plotting script and the requested PNG/PDF/SVG figure by default. Use for publication-style plot scripts, science/nature figures, plot-from-data scaffolds, and Korean or English requests for matplotlib graph scripts.
In this repository38 other files it gives its agents.
- academic-cvSkill
- academic-jobsSkill
- academic-proseSkill
- adversarial-reviewSkill
- bibtex-genSkill
- chrome-sessionSkill
- clickup-cliSkill
- clickup-todaySkill
- commit-triageSkill
- concept-explainerSkill
- contact-emailSkill
- cover-letterSkill
- dropboxSkill
- handdrawn-schematicSkill
- hep-rumor-millSkill
- journal-club-reviewSkill
- magi-researchSkill
- md2pdf-typoraSkill
- morgenSkill
- overleaf-section-workflowSkill
- overleapSkill
- peer-agent-reviewSkill
- proton-mailSkill
- reference-searchSkill
- remarkable-syncSkill
- research-backupSkill
- research-boardSkill
- research-logSkill
- research-portalSkill
- research-proposalSkill
- research-reportSkill
- research-statementSkill
- vastaiSkill
- wide-slide-illustratorSkill
- workshop-paper-reviewSkill
- xkcd-pySkill
- zai-web-searchSkill
- zoom-summarySkill
The file
We split this skill into the rules it gives an agent
6 rules. About 0 are common, 2 are distinctive, and 4 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 29
Every generated script MUST keep these load-bearing patterns intact.
- Very specific: line 47
Default filename is plot.png.
- Very specific: line 71
Write the .py file with the Write tool.
- Very specific: line 75
Prefer uv run <path> when the current project is the active uv project.
- Very specific: line 81
Verify that every requested output file exists and is non-empty.
- Distinctive: line 89
If the user explicitly requests script-only, stop after step 5 and return an execution hint instead of running the script.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
scienceplot-py — Matplotlib Plot + Figure Generator (scienceplots / science+nature)
Generate a Python plotting script that always follows the user's lab
template shape, execute it, and verify the rendered figure artifact. The skill
returns both the .py path and the generated figure path(s). The default
workflow is script plus figure generation; use script-only mode only when the
user explicitly requests a script without execution.
The canonical template lives at
~/Socialst/Templates/PyPlot_Template/pq_plot.py. Every variant in this
skill is a structural extension of that file.
Mandatory style invariants
Every generated script MUST keep these load-bearing patterns intact. Do not "clean up" any of them, even if they look redundant or unused.
import scienceplots— required. It registers thescienceandnaturestyles by import side-effect; the symbol is never referenced directly. Linters will flag it as unused — keep it anyway.- Style context block — all plotting code lives inside
with plt.style.context(["science", "nature"]):. Never substituteplt.style.use(...)or call any plotting outside this block. pparamdict — axis configuration is a dict, applied withax.set(**pparam). Do not inlineax.set_xlabel(...),set_ylabel(...),set_title(...)calls whenpparamwould do.ax.autoscale(tight=True)— called on every axis, beforeax.set(**pparam). For subplots, loop over all axes.- Raw-string LaTeX — every label, title, legend entry uses
r'...'. Non-raw'$x$'works for the literal$x$but breaks the moment a backslash appears (\alpha,\sigma,\mathrm{...}). Alwaysr-prefix. - savefig —
fig.savefig(<path>, dpi=300, bbox_inches='tight'). Default filename isplot.png. PDF/SVG output is fine if asked, but keepdpi=300andbbox_inches='tight'. (Bump higher only if the user explicitly asks for it — the lab default is 300.)
Workflow
- Gather what the user has not already given:
- Data source: parquet / CSV /
.npy/.npz, plus the file path - x / y column or array names (and y-error if errorbar)
- Plot variant: single line, multi-line + legend, scatter / errorbar, or subplots (rows × cols)
- Axis labels, title, legend labels (LaTeX OK — strings will be wrapped in raw-string form)
- Output path (default:
plot.pngnext to the data file or in cwd)
- Data source: parquet / CSV /
- Pick the matching template under
references/:single_line.py— one series, one ax (mirrorspq_plot.py)multi_line.py— multiple series, one ax, with legendscatter_errorbar.py—ax.errorbar(...)(orax.scatter(...))subplots.py— multi-panelfig, axes = plt.subplots(rows, cols)
- Swap in the requested data-loader block. See
references/data_loaders.mdfor parquet / CSV /.npy/.npzsnippets. - Substitute column names, label strings, title, output filename. Preserve every invariant from the section above.
- Write the
.pyfile with the Write tool. Preserve the requested output location in the script rather than relying on an accidental current working directory. - Render the figure by executing the generated script from the project root.
Prefer
uv run <path>when the current project is the active uv project; when the project environment is in a subdirectory, useuv run --project <project-dir> python <script-path>. Do not install packages automatically: ifscienceplots, matplotlib, pandas, or numpy is missing, report the dependency error and the command needed after the user fixes the environment. - Verify that every requested output file exists and is non-empty. For report
figures, generate the requested raster output (normally PNG) and a PDF or
SVG companion, each with
dpi=300andbbox_inches='tight'. If the figure is available to the harness, inspect it for clipped labels, missing glyphs, empty axes, and accidental transparent backgrounds. - Return the script path, generated figure path(s), the exact render command, and a short render-status summary.
If the user explicitly requests script-only, stop after step 5 and return an execution hint instead of running the script.
Data sources
| Source | Imports | Read call |
| Parquet | import pandas as pd | df = pd.read_parquet('data.parquet') |
| CSV | import pandas as pd | df = pd.read_csv('data.csv') |
NumPy .npy | import numpy as np | data = np.load('data.npy') (then column-slice) |
NumPy .npz | import numpy as np | data = np.load('data.npz'); x = data['x'] |
Full snippets in references/data_loaders.md.
Templates
Each reference is a self-contained, runnable skeleton matching the mandatory style invariants. Read the file, adapt strings/columns in memory, then Write the result to the user's chosen path.
| Variant | File |
| Single line (base) | references/single_line.py |
| Multi-line + legend | references/multi_line.py |
| Scatter / errorbar | references/scatter_errorbar.py |
| Subplots (multi-panel) | references/subplots.py |
| Data loaders | references/data_loaders.md |
What this skill does NOT do
- It does not install
scienceplots/matplotlib/pandas/numpy. Assume they are already available in the target environment; dependency installation remains the user's or project's responsibility. - It does not produce methodology / architecture / pipeline diagrams. For
those, use
wide-slide-illustratororhanddrawn-schematic. - It does not change the style invariants. If the user asks for a different
style (e.g.,
seaborn,ggplot, plain matplotlib), tell them this skill is specifically for the science+nature template and offer to write the alternative as a plain script outside the skill. - It does not silently treat a successful Python exit as proof of a valid figure: output existence and basic visual sanity checks are part of the rendering workflow.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

