Round-12 review found two artifact-claim gaps: --all never touched
receipts, and receipt mode checked fingerprint/leaf_hash only when
present. Now:
- --all enumerates receipts/*.receipt.json and runs the full binding
list on each (type tag, STH signature, REQUIRED key fingerprint,
log_id vs log-metadata, STH membership in the published history,
REQUIRED leaf_hash vs the named entry, tree_size agreement, receipt
hash_algorithm + log_id consistency, inclusion proof).
- --receipt FILE routes through the same function.
- Malformed values (bad hex, missing sizes) are failures, not crashes.
- NEW verify_selftest.py: 11-case adversarial battery (mutated real
receipts must be REJECTED: no fingerprint, no leaf_hash, wrong type,
forged root, size mismatch, wrong log_id; honest controls pass;
structural-only never claims full; no-openssl exits 2). GREEN.
One bug caught by the honest controls during development: the new
hash-algorithm check assumed 'sha256' but deployed heads carry
'RFC9162_SHA256' - fixed against reality, plus receipt-level
hash_algorithm/log_id binding added.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
GPT-5.6 live-site review + Fable live verification:
- verify.py §4: was fail-OPEN — no openssl ⇒ signatures 'skipped' ⇒
still 'OK'. Now fail-CLOSED (exit 2 unless --structural-only, which is
explicitly labelled a reduced check). --receipt was Merkle-path-only;
now verifies the receipt's STH signature, key fingerprint, log_id,
tree_size agreement, leaf_hash-vs-entry, and history membership before
the inclusion proof. --all now also checks latest-sth == final history
head and size == leaf count, and constant log_id. Adversarially tested:
fail-closed exit 2; forged unsigned-root receipt REJECTED (was the
'inclusion VALID' hole); honest receipt + full run still exit 0.
- README §8: I had written 'the first deployed transparency log to carry
proofs of its own honesty' this session — an overclaim (the corpus
proves properties of the accumulator MODEL, not operator honesty).
Now: 'kernel-checked proofs about the accumulator model underlying its
inclusion/consistency reasoning', 'first' softened to 'unaware of a
precedent'.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Broadened 'Rust libraries' framing to 'specific software'; noted the
first-of-its-kind self-attestation and the current 13-leaf head.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The log now carries kernel-checked proofs of its own accumulator
machinery. Entry 13 attests the ltl-accumulator-verified corpus at
freeze 172a1d0: 61 certificates, all proven with exact axiom cones,
single opaque-SHA-256 axiom boundary. Scope (in the leaf): the
mechanized recursive model is kernel-checked; deployed-verifier
correspondence is finite differential testing; the deployed
consistency verifier is not extensionally equal to the model and
relies on an unmechanized authentic-size/root invariant (gaps 14/15).
tree 12 -> 13, root 3488a2d0; head signed verified-dalek-serial.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The public face of the Lean Transparency Log: every leaf (including the
first run's honest audit-failure attestations), the full STH history
(witness channel), per-component receipts, the provider public key, and
a standalone stdlib-only verify.py.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
