Skip to content

ArgLib Documentation

ArgLib is a Python library for creating, importing, analyzing, and reasoning over warrant-gated argument graphs. The library is currently in early development, but the core graph model, scoring, diagnostics, and CLI are ready to use.

What you can do today

  • Build argument graphs with typed nodes and relations.
  • Export graphs to JSON or Graphviz DOT.
  • Run warrant-gated credibility propagation and diagnostics.
  • Use a lightweight CLI for DOT export, validation, and diagnostics.
  • Define argument bundles (arguments as subgraphs) and project to higher-level graphs.
  • Attach evidence cards/supporting documents and propagate credibility scores.
  • Run deterministic evidence scoring and edge validation helpers.

Quickstart

uv sync
uv run pytest

Example

from arglib.core import ArgumentGraph
from arglib.reasoning import compute_credibility

graph = ArgumentGraph.new(title="Parks")
c1 = graph.add_claim("Green spaces reduce urban heat.", type="fact")
c2 = graph.add_claim("Cities should fund parks.", type="policy")
graph.add_support(c1, c2, rationale="Cooling improves health", gate_mode="OR")

credibility = compute_credibility(graph)
scores = credibility.final_scores