Provenance

Every CognitionFM artifact is listed here with everything needed to regenerate and verify it: recipe, seed, duration, engine version and commit, and the SHA-256 checksum of the exact file.

Renders are deterministic. python -m cognitionfm render --recipe <recipe> --duration <duration> --seed <seed> at the listed engine commit regenerates the audio byte-for-byte on the same platform and library versions (compare with shasum -a 256); on different hardware, floating-point rounding can shift the checksum without audibly changing the mix, so the recipe, seed, and commit remain the canonical definition of each artifact. Cover art derives from the same seed. This ledger is the ownership record: nothing published samples, remixes, or launders anyone else's work. Status is local-test until an artifact is actually published, at which point its public URL is added. Rows marked site-preview are the lossy AAC excerpts playable on the landing page, derived from the listed masters; their checksums cover the encoded file as served.