CascadeLens v0.6.0

CascadeLens v0.6.0 expands the executable decision-case library from 12 to 16 cases and adds a shared 15-capability taxonomy, a generated capability matrix, a 25-run normalized-severity sensitivity surface, and evidence-bounded decision-brief export.

Verified release scope:

• 16 executable scenario-only cases
• 11 bounded connector contracts
• 4 frozen official-source snapshots with 4,027 normalized facts
• 222 BEA sector-level dependency edges, explicitly MODEL_INFERRED and upper-bound-only
• 133 Node tests and 23 Python tests
• clean-archive verification and byte-reproducible offline build

Evidence boundary:
This is software and release verification, not empirical validation. Historically scored cases, external method/domain reviews, structured user studies, verified organizational adoptions, and demonstrated real-world impacts all remain 0.

CascadeLens v0.5.1

Clearer public workflow

• Added a compact decision-to-RiskPack overview and retained the complete decision and verification graph in an expandable technical section.
• Aligned the Python package, website, active install commands, citation, evidence ledger, issue forms, and CycloneDX SBOM to v0.5.1.
• Updated the pinned CodeQL action to the reviewed v4.37.7 tag commit.

Verified release

• 23 Python tests passed locally and on the Python 3.11–3.13 GitHub matrix.
• 130 TypeScript unit and contract tests, 8 documentation tests, 8 rendered-route tests, and 2 accessibility tests passed.
• All 12 scenario-only cases and four frozen public snapshots revalidated.
• Dependency audit returned 0 vulnerabilities; security and 20,000-node performance gates passed.
• Linux/macOS CI and detached no-Git verification reproduced the complete production build digest exactly.

Evidence boundary

This patch changes presentation and release identity only. The machine evidence ledger remains at 0 historically scored cases, 0 external reviews, 0 structured user studies, 0 verified organizational adoptions, and 0 demonstrated real-world impacts. All 12 public cases remain scenario_only.

CascadeLens v0.5.0 — Evidence-Gated Relational Data

What changed

• Added the official 2023 BEA Commodity-by-Industry Direct Requirements connector and a frozen, hash-pinned source snapshot.
• Expanded the public-data catalog to 11 connectors, 4 frozen official-source runs, 4,027 normalized facts, and 222 positive sector-level inputs_to edges.
• Kept every BEA edge model-inferred and upper-bound only; these edges are not firm-level links, causal estimates, or eligible primary-estimate evidence.
• Added a machine-validated evidence ledger plus ready-to-run protocols for historical replay, external review, structured user study, adoption, and impact.
• Hardened XLSX ingestion against unsafe paths, archive bombs, formulas, duplicate cells/codes, unexpected dimensions, and invalid coefficients.

Evidence boundary

The 12 demonstration cases remain scenario_only. Accepted counts for historical scored cases, external method/domain review, structured user research, organizational adoption, and real-world impact remain 0 until qualifying public evidence exists. Internal tests, repository stars, templates, and public hosting do not increment those counts.

Verification

• Node tests: 130/130
• Python tests: 23/23
• Rendered routes: 8/8
• Accessibility checks: 2/2
• Security findings: 0
• Dependency audit vulnerabilities: 0
• Fresh no-Git archive install and full CI: pass
• Two offline production builds reproduced byte-for-byte: e93bf7e2645607d1668f6e5fe2096c35812a93d5dbba70e5d0821960e788ae7a

See checksums.sha256, release-manifest.json, and verification-report.json for release integrity. This release verifies software behavior and artifact integrity only; it does not establish external validation, adoption, or real-world impact.

CascadeLens v0.4.0 — Python-first analysis

Python-first analyst workflow

• Install a dependency-free Python 3.11+ package from this tag.
• Import JSON, edge-list CSV, GraphML, or NetworkX and run bounded cascade analysis through the CLI, API, or Jupyter.
• Create and independently recompute a checksummed Python RiskPack.
• Match Python lower, central, upper, intervention, and benchmark outputs against all 12 reviewed browser cases.

Workbench and documentation

• Explore all 12 scenario-only cases in the hosted Workbench.
• Import graph and ShockScript files locally in the browser; files are not uploaded.
• Follow a real captured workflow GIF, three Python tutorials, checked examples, and a formal methods note.
• Use stricter duplicate, numeric-bound, GraphML entity, directionality, symlink, and overwrite guards.

Verification

• Python 3.11, 3.12, and 3.13 CI passed.
• Complete web CI, CodeQL, dependency audit, accessibility, and the 20,000-node engineering budget passed.
• Ubuntu and macOS reproduced the complete offline web build.
• The attached no-.git verification receipt reran Python, full web CI, deterministic artifact generation, and two offline builds from the source archive.

Install

pip install "cascadelens @ git+https://github.com/limingrui679-design/CascadeLens.git@v0.4.0"
cascadelens demo --out demo-riskpack

Evidence boundary

This release contains 0 historically scored cases, 0 external validations, 0 verified organizational adoptions, and 0 demonstrated real-world impacts. The 12 cases are deterministic scenario_only research demonstrations, not forecasts or client projects.

See the v0.4.0 five-pass review and adoption/validation ledger.

CascadeLens v0.3.2

CascadeLens v0.3.2

This immutable release closes the independently reproduced RiskPack, reproducible-build, adapter-order, and style-CSP findings, plus the stale-output gap detected by detached pre-publication verification.

Assurance

• 123/123 unit and integration tests pass, with 0 failures, skips, or todos.
• Full CI and CodeQL pass on the exact tagged commit.
• Offline production builds reproduce exactly on Ubuntu and macOS.
• A fresh source archive without .git installs, regenerates artifacts, passes the complete CI suite, and reproduces the exact production build digest twice.
• The release includes source ZIP/tarball, CycloneDX SBOM, canonical relative checksums, exact release manifest, and detached verification receipt.
• Repository-level immutable releases lock the published tag, commit binding, assets, and GitHub release attestation.

Evidence boundary

All 12 reference cases remain scenario_only. This release supplies 0 historically scored cases, 0 external method or security validations, and 0 claims of organizational adoption or demonstrated real-world impact.

CascadeLens v0.3.1

CascadeLens v0.3.1

This release closes the independently reproduced RiskPack, reproducible-build, adapter-order, and style-CSP findings, plus the stale-output gap detected by the detached pre-publication verifier.

Assurance

• 123/123 unit and integration tests pass, with 0 failures, skips, or todos.
• Full CI and CodeQL pass on the exact tagged commit.
• Offline production builds reproduce exactly on Ubuntu and macOS.
• A fresh source archive without .git installs, regenerates artifacts, passes the complete CI suite, and reproduces the exact production build digest twice.
• The release includes source ZIP/tarball, CycloneDX SBOM, canonical relative checksums, exact release manifest, and detached verification receipt.

Evidence boundary

All 12 reference cases remain scenario_only. This release supplies 0 historically scored cases, 0 external method or security validations, and 0 claims of organizational adoption or demonstrated real-world impact.

CascadeLens v0.2.1

Audit-remediation release with deterministic RiskPack recomputation, locale-independent canonical digests, time-correct replay and intervention semantics, end-to-end connector evidence, hardened archive verification, and machine-readable hosted build identity.

Verification: 95/95 tests, fresh no-.git rebuild, deterministic regeneration, CycloneDX SBOM, archive budgets, 0 known dependency vulnerabilities, and enforced performance budgets.

Evidence boundary: all 12 reference cases remain scenario_only; this release contains 0 historically scored cases, 0 external validations, and 0 claims of organizational adoption or demonstrated real-world impact.

CascadeLens v0.1.1

CascadeLens v0.1.1

This patch hardens replay scoring, public response security, accessibility coverage, and release assurance.

Fixed

• Block unknown outcome nodes, duplicate node observations, and malformed outcome timestamps.
• Add same-origin resource isolation and a two-year HTTPS transport policy.
• Verify branded 404 behavior and audit every dynamic case page for detectable accessibility violations.
• Update the compatible Cloudflare build and deployment toolchain.

Verification

The attached source archive was extracted without .git, installed from the lockfile, regenerated deterministically, and passed the complete CI gate: 71 core tests, rendered-route checks, all case-page accessibility checks, security audit, and the 20,000-node performance profile. See verification-report.json and checksums.sha256.

This release still contains 12 scenario-only cases and 0 historically scored cases. Software verification and public hosting do not establish empirical model validity, external review, adoption, or real-world impact.

CascadeLens v0.1.0

CascadeLens v0.1.0 is the first software-verified research release of the evidence-graded, bitemporal world graph and executable shock-analysis platform.

Release scope:

• 10 bounded official/public-source connector contracts.
• 12 deterministic scenario-only reference cases with verified RiskPacks.
• WorldGraph, ShockScript, cascade bounds, InterventionLab, Observability Frontier, CascadeBench, CLI, SDK, and public web product.
• Fresh-archive install, deterministic rebuild, full CI, accessibility, security, dependency-audit, and 20,000-node performance gates.

Evidence boundary: this release contains 0 historically scored cases, 0 external validations, and 0 claims of organizational adoption or real-user impact. Software verification and public hosting do not establish empirical predictive validity.

Public product: https://cascadelens.limingrui2.chatgpt.site

Verify downloads against checksums.sha256; verify the fresh-archive receipt against verification-report.sha256.