verification architecture
ai generates claims faster
than anyone can check them.
manifold control designs the systems that keep the volume honest: registries that give every definition one identity, provenance that records what was checked and by which verifier, and proof-term evidence that shows what a kernel actually accepted. built for organizations deploying ai where trust, verification, and decision support have to survive scale.
live work
running systems, rebuilt from source in ci. this section grows as new work ships.
csr: live registries live
a corpus semantic registry: documents define, the registry identifies. three browsable registries: a real mathematics preprint's claim table (19 symbols, 8 proved, 4 open, sha256-pinned against the pdf) plus two worked examples. edit a pinned source and every dependent claim is flagged in ci until re-verified.
a paper that audits itself live
the claim-status table of a singular fold model for capacity-constrained dynamics (v67), maintained machine-readably: per-claim verification state, dependency edges between theorems, and drift detection against the pdf. what "the paper says X" looks like when it's checkable.
the verification stack
open source, mit licensed, each piece usable on its own. identity, provenance, and evidence: the three layers an audit trail needs.
#introspect: proof-term dag + leakage report (sorry, mvars, dependency surface)
three rules that hold throughout
the discipline underneath all of it.
events record standing decisions. the deciding happens elsewhere.
events reference identities. the registry mints them.
credit derives from events. the flow runs one way.
about
manifold control is the practice of james kovalenko (charlottesville, va): verification architecture and practical ai deployment: systems for trust, verification, and decision support as ai increases the volume of information that must be reviewed.
the public work above is the method in miniature: make claims addressable, pin what they depend on, and let a build step say what still holds. the same architecture applies to spec-driven codebases, research corpora, compliance documentation, and ai agent workflows.