World graph → shock → decision evidence
CascadeLens compiles facts, assumptions, missing links, and decisions into one auditable chain—then keeps each claim in its proper evidence class.
Verified build artifact
Context citation only. Topology and numbers are declared assumptions; zero launch cases claim empirical scoring.
Python-first · one-line install
Import JSON, CSV, GraphML, or NetworkX. Generate bounded results and a recomputable RiskPack locally.
pip install "cascadelens @ git+https://github.com/limingrui679-design/CascadeLens.git@v0.5.0"cascadelens demo --out demo-riskpackOne chain of custody
The platform is designed to stop when evidence, licensing, temporal validity, or feasibility is insufficient.
Bitemporal nodes, edges, sources, license modes, and content-addressed snapshots.
A strict YAML/JSON contract for targets, operations, horizons, assumptions, and constraints.
Lower, central, and upper propagation expose missing relationships instead of smoothing them away.
Ranks which missing fact could change the decision—without asserting that the fact is true.
Inputs, results, and checksums backed by deterministic derived-output recomputation.
Activation-dated bundles, horizon-specific Pareto trade-offs, and evidence-required status.
Truthful by construction
Extracted and inferred relations are barred from primary estimates. A context page cannot substitute for a preserved, hashed data artifact.
Reproduce the whole claim