matplotlibapi

name	matplotlibapi
description	Profile local data, explain chart recommendations, repair and validate PlotSpecs, and render accessible charts through MatplotLibAPI's schema-driven contract.

MatplotLibAPI

Credential policy

MatplotLibAPI requires no credentials. Never request API keys, tokens, hosted authentication, secret-store access, or remote data uploads. Process only local files and table data the user has explicitly supplied.

An optional .env may contain only non-secret local runtime settings:

MPLBACKEND=Agg

Setup

python -m pip install --upgrade "git+https://github.com/fatmambot33/MatplotLibAPI.git"

Verify the installation:

matplotlibapi doctor
matplotlibapi test
matplotlibapi plugins conform

Canonical workflow

1. Profile data locally with matplotlibapi inspect or profile_dataframe.
2. Review ranked, explainable chart options with matplotlibapi recommend or recommend_plots.
3. Discover canonical plot descriptors with matplotlibapi plots list or create_registry().context.list_descriptors().
4. Build a PlotSpec using semantic column roles in encoding, visual settings in options, and accessibility or number-format choices in presentation.
5. Use matplotlibapi repair or suggest_plot_spec_repairs for opt-in, structured corrections. Never silently mutate a user specification.
6. Validate before execution.
7. Render through execute_plot or the CLI. Do not bypass the canonical executor unless the registry cannot express the requested result.
8. Keep input and output paths inside an explicit workspace.
9. Report the output path, profile truncation status, recommendation reasons, alt text, and any structured warnings.

Discovery and migration

from MatplotLibAPI import (
    PlotSpec,
    create_registry,
    openai_tool_definitions,
    v5_compatibility_status,
)

registry = create_registry()
print(registry.context.list_descriptors())
print(PlotSpec.json_schema())
print(openai_tool_definitions(registry=registry))
print(v5_compatibility_status())

Use timeseries as the canonical chart name. Treat timeserie, histogram, and pie as compatibility aliases and use migration diagnostics before 5.0. The breaking-removal gate must remain closed before its documented date.

Use Matplotlib for static output and Plotly only when interactivity materially improves the result. Prefer deterministic, reproducible transformations and preserve source data.