Forensic CPM Engine — Reproducibility Certificate

One-page expert-disclosure attachment. The engine that produced your forensic report is independently reproducible by any opposing expert.

VERSION 2.9.16 · SHA-256-LOCKED · SIGSTORE-SIGNED

Engine Identity

JavaScript engine SHA-256

4815ba956d733ea4c008da4337bf2fa36c7d9ef3630a027475e28a4d2bea5e0d

Python reference SHA-256

1537dfd490062bfe1b726f86cabf4731d8a2218b6013732c8a53256a2bcb7e7c

GitHub commit (tag v2.9.16)

c0af428c9b6465f073f5d00d2d0f4a80507cc69f

Topology-hash format

sha256-canonical-v2 (prefix v2:)

Four Public Deployment Surfaces — Bit-Identical

Surface	Install Path	Verification
npm package	npm install cpp-cpm-engine@2.9.16	npm view cpp-cpm-engine@2.9.16 dist.shasum
GitHub release	github.com/danafitkowski/cpp-cpm-engine/releases/tag/v2.9.16	gh attestation verify cpp-cpm-engine-2.9.16.tgz
Railway-hosted MCP	curl mcp.criticalpathpartners.ca/cpm-engine.js	sha256sum — compare to cell above
Local skill bundle	~/.claude/skills/cpp-forensic-mcp/static/cpm-engine.js	sha256sum — compare to cell above
MCP registry index	registry.modelcontextprotocol.io/v0/servers?search=cpp-cpm	version === 2.9.16 (later builds have since shipped)

Reproduction in under a minute, zero npm dependencies, any opposing expert

git clone https://github.com/danafitkowski/cpp-cpm-engine
cd cpp-cpm-engine && git checkout v2.9.16
npm test                # 878 unit tests
node cpm-engine.crossval.js   # 43 fixtures × 444 checks JS↔Python parity

Verifiable Evidence

Unit tests passing

878 / 878 assertions

Cross-validation

43 fixtures · 444 / 444 checks

CI matrix

3 OS × 3 Node = 9 cells per push

Engine lines of code

JS 6,950 · Python 1,304

Holiday calendars

66 jurisdictions (CA+13, US+50, +UK)

Sigstore signature

Rekor public transparency log

AACE-Canonical Method Identifiers

AACE 29R-03 — Forensic Schedule Analysis. MIP 3.3 Observational/Dynamic/Contemporaneous As-Is (Windows); MIP 3.6 Modeled/Additive/Single Base (Impacted As-Planned); MIP 3.7 Modeled/Additive/Multiple Base (TIA); MIP 3.8 Modeled/Subtractive/Single Simulation (Collapsed As-Built).
AACE 49R-06 — Critical Path identification methods (LPM · TFM · MFP).
AACE 52R-06 — Time Impact Analysis (fragnet insertion).
DCMA 14-Point Assessment — full coverage in CI.
SCL Delay & Disruption Protocol, 2nd Edition (2017) — methodology alignment.

Daubert Framework Compliance

FRE 702 as amended December 1, 2023 — methodology reliability (Rule 702(c) and 702(d)) demonstrated to preponderance of evidence. All four Daubert prongs addressed in buildDaubertDisclosure().
Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993) — foundational reliability framework: testability, peer review, error rate, general acceptance.
FRCP 26(a)(2)(B) — expert disclosure of methods and data. Engine version + topology hash + manifest fields satisfy disclosure.
Proposed FRE 707 — would apply Rule 702(a)–(d) reliability standards to machine-generated evidence offered without an expert witness. That is squarely what this engine produces, so its output must be able to show sufficient input data, reliable principles and methods, and reliable application to the facts. This certificate exists to make that showable. Status: proposed only. The US Courts list it as projected to take effect 1 December 2027 and note that proposed amendments may be delayed or withdrawn. Pending amendments status. Not to be confused with FRE 706, which governs court-appointed experts.