Legitimacy v1.0.0

Legitimacy v1.0.0

The first tagged release of Legitimacy packages a machine-checked governance
kernel for agentic AI systems, a verified impossibility result for scarce
peer-relative allocation, and capability-scaling bounds that make the resulting
stability cliff explicit.

Highlights

• A Lean 4 LegitimacyKernel with five explicit obligations:
  certifiability, observability, corrigibility, compositional safety, and
  non-vacuity.
• A machine-checked forcing theorem for reachable scarce, structurally
  peer-relative decision stages.
• A shared theorem carrying the forced sacrifice and capability cliff on the
  canonical claimant-interaction lift.
• Executable Rust audits over graph fixtures extracted from agent-framework
  governance surfaces.
• Reproducible verification and publication tooling, with zero sorry,
  admit, or first-party axiom in the Lean substrate.

Release assets

The release includes the two primary manuscripts and their verification
records:

• 02-semantic-legitimacy-kernels.pdf
• 03-impossibility-theorem.pdf
• CONTENT-CHECKSUMS
• BUILD-PROVENANCE.txt
• GITHUB-RELEASE-SHA256SUMS

Paper 04, Capability Scaling, Spectral Stress, and Stability Cliffs, is
included as source in the repository. Papers 02–04 are being prepared for a
separate archival submission.

Verify the attached payload from the directory containing the downloaded files:

sha256sum -c GITHUB-RELEASE-SHA256SUMS

The complete verification gate is:

bash scripts/verify.sh --release-gate