Nineteen crates, 613,565 lines of first-party Rust: dense and sparse linear algebra, integration, interpolation, optimization, FFT, signal, spatial, special functions, statistics, clustering, ndimage, IO, constants, plus a runtime crate holding the portfolio engine and a conformance crate holding the differential harness. 10,174 inline tests, 795 conformance test files, 16 Python oracle scripts, 96 fuzz targets, 18 canonical parity packets. Zero unsafe blocks under a workspace-wide forbid. A real release: v0.2.0, git tag plus GitHub release, all 19 crates published on crates.io.
Three findings matter. First, CASP is a real idea. The Condition-Aware Solver Portfolio probes matrix conditioning, computes a posterior over condition states with a conformal calibrator, minimizes expected loss over solver actions against a calibrated loss matrix (the matrix values are verified in code, a real const fn), and returns a synchronous audit certificate. It is wired into five numerical domains, each entry point verified calling the selection routine. The calibration quality is maintainer-asserted, never independently validated. But the architecture is real.
Second, the scorecard disavows itself. The release scorecard file opens by telling you that not one figure in it clears the project's evidence gate. The FAQ refuses a speed headline: "For individual kernels, mileage varies by problem size, conditioning, and structure." For a "reimplementation of SciPy," the absence of quotable head-to-head numbers is a thesis-level gap, honestly disclosed.
Third, the parity census is exactly what it says it is, and the project says so. 1,300 of 1,300 routines are name-matched. The README admits the per-routine list of what each differential test actually asserts "does not exist yet" and that "referenced is weaker than compared." Parity here is a naming convention until the assertion census gets built.
Uncomfortable finding
Bus factor (the number of key contributors who could disappear before the project stalls) 1 with AI-authored provenance: 6,395 commits carry the claude contributor identity against 158 from the human owner. crates.io downloads sit at 11–45 per crate. The license is MIT plus the OpenAI/Anthropic rider, not OSI open source, and it denies even benchmarking and analysis to the two labs themselves, the parties most able to validate it independently; people who run those labs' agents, the project's named audience, are not barred. Zero independent benchmarks, reviews, or deployments were found. The README drifts against the tree in a dozen places, including one outright contradiction (one section says "not yet published to crates.io" while another says published; crates.io confirms published). The README badge implying asupersync is the async runtime is disproven: asupersync appears only in the conformance crate for evidence-pack encoding, never as a runtime. source The rider also bars Anthropic from training on this code, which Claude-family models helped write; that restricts the lab, not the maintainer's own use of its models.
Corrected 2026-09-24: an earlier version called this a tension; the rider bars the two labs and anyone acting for them, not people who use their models (changelog).