verification: bind the statements, the specifications, and the model (P1-a)
Phases 3/3b establish what each certificate RESTS ON. Neither says what it
SAYS, nor what it is ABOUT. A certificate gutted to a tautology of the same
axiom cone passes both; so does one whose reference definition has been
redefined to BE the extracted code, at which point the theorem reads
`loop = loop` and every cone is byte-identical.
Phase 3c closes that. Proofs/Audit.lean emits a canonical block holding the
policy constants, every certificate's fully-elaborated statement (pp.all, so
implicit arguments, instances and universe levels are visible), and the body
of every specification constant transitively reachable from those statements.
Its SHA-256 is pinned in check.sh and the block itself is committed as
AUDIT-MANIFEST.txt, so a mismatch is DIFFED, not merely reported. 31
certificates, 68 specification constants per repository.
Two tiers, not one. These forks have an arithmetic tier that must stay
oracle-free and an apex tier carrying this fork's hash and wire-format axioms,
and the apex boundary genuinely differs per fork (dalek 8 extra names, anza 4,
risc0 and betrusted 5). One shared constant would have widened the arithmetic
tier to accept hash oracles, which is the most valuable property these repos
have. Each auditor is generated from its own repository's policy.
Phase 0b pins the extracted model. This was not a precaution: risc0 and
betrusted were observed emitting BYTE-IDENTICAL audit-manifest digests
(6c821b8e…) while shipping demonstrably different extracted models, their
point-doubling routines differing in operation order. A statement names an
extracted function; it does not contain that function's body. Binding
statements is not binding the subject. Membership derives from the filesystem,
so a new model file fails closed.
selftest-statements.sh attacks both phases with ten cases, each asserting a
specific diagnostic: an edited model body, an unlisted model file, a widened
policy, a hand-edited committed block, a certificate dropped from the auditor
WITH the digest refreshed to match, and a gutted statement whose cone is
unchanged. It lifts the phases out of check.sh at run time, so it attacks the
shipping gate rather than a copy.
Two bugs found and fixed during that testing, both mine: Phase 3c read `$0`
after `cd "$AENEAS_LEAN"`, and $0 is the caller's relative path; and the
axgate self-test compared the tree against a pristine checkout rather than
against how it found it. A third expectation was wrong rather than the code —
widening the apex boundary is caught by the exact-cone requirement before the
digest ever runs, which is a stronger rejection, and the test now says so.
All sixteen runs green at these commits: four main buttons, four axgate
self-tests, four binding self-tests, four scalar buttons.
TRUSTED-BASE.md records what this binds and, at equal length, what it does
not: a digest binds identity, not meaning; an author can rotate the pins in
one commit and is caught by review, not by the script; and pinning the model
says nothing about whether Charon and Aeneas translated the Rust faithfully.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-07-28 22:38:22 +00:00
|
|
|
|
#!/usr/bin/env bash
|
|
|
|
|
|
# ─────────────────────────────────────────────────────────────────────────────
|
|
|
|
|
|
# selftest-statements.sh — adversarial self-test for the BINDING phases,
|
|
|
|
|
|
# check.sh 0b (the extracted model) and 3c (statements and specifications).
|
|
|
|
|
|
#
|
|
|
|
|
|
# These phases exist because Phases 2b/3/3b establish what a certificate RESTS
|
|
|
|
|
|
# ON and never what it SAYS, nor what it is ABOUT. This script performs the
|
|
|
|
|
|
# attacks that move no axiom cone at all, and asserts each is rejected FOR THE
|
|
|
|
|
|
# STATED REASON.
|
|
|
|
|
|
#
|
|
|
|
|
|
# It lifts Phase 3c out of check.sh at run time, so it attacks the shipping
|
|
|
|
|
|
# gate rather than a copy that can drift away from it.
|
|
|
|
|
|
#
|
|
|
|
|
|
# Requires a prior green build (Proofs/*.olean present). The cheap cases each
|
|
|
|
|
|
# recompile only Proofs/Audit.lean (~10 s). The gutted-statement case also
|
|
|
|
|
|
# recompiles one certificate module and is therefore slower; skip it with
|
|
|
|
|
|
# SKIP_SLOW=1 if you only want the fast gates exercised.
|
|
|
|
|
|
# ─────────────────────────────────────────────────────────────────────────────
|
|
|
|
|
|
set -uo pipefail
|
|
|
|
|
|
source ~/aeneas-toolchain/env.sh
|
|
|
|
|
|
HERE="$(cd "$(dirname "$0")" && pwd)"
|
|
|
|
|
|
AENEAS_LEAN="$AENEAS_HOME/backends/lean"
|
|
|
|
|
|
TIMEOUT="${LEAN_TIMEOUT:-900}"
|
|
|
|
|
|
export LEAN_MEM_MB="${LEAN_MEM_MB:-8192}"
|
|
|
|
|
|
CORES="${LEAN_MAX_CORES:-0-3}"
|
|
|
|
|
|
SKIP_SLOW="${SKIP_SLOW:-0}"
|
|
|
|
|
|
|
|
|
|
|
|
STASH="$(mktemp -d)"
|
|
|
|
|
|
FAILURES=0
|
|
|
|
|
|
# Recorded before anything is touched, so the restore check compares against
|
|
|
|
|
|
# reality rather than assuming a pristine checkout.
|
|
|
|
|
|
TREE_AT_START="$(cd "$(dirname "$0")/.." && git status --porcelain)"
|
|
|
|
|
|
|
|
|
|
|
|
cleanup() {
|
|
|
|
|
|
[ -f "$STASH/Audit.lean" ] && cp "$STASH/Audit.lean" "$HERE/Proofs/Audit.lean"
|
|
|
|
|
|
[ -f "$STASH/AUDIT-MANIFEST.txt" ] && cp "$STASH/AUDIT-MANIFEST.txt" "$HERE/AUDIT-MANIFEST.txt"
|
|
|
|
|
|
[ -f "$STASH/check.sh" ] && cp "$STASH/check.sh" "$HERE/check.sh"
|
|
|
|
|
|
[ -n "${GUT_MOD:-}" ] && [ -f "$STASH/gut.lean" ] && cp "$STASH/gut.lean" "$HERE/Proofs/$GUT_MOD.lean"
|
|
|
|
|
|
[ -n "${GENF:-}" ] && [ -f "$STASH/genfile.bak" ] && cp "$STASH/genfile.bak" "$HERE/gen/$GENF"
|
|
|
|
|
|
[ -n "${GENF:-}" ] && rm -f "$HERE/gen/$(dirname "$GENF")/ZZExtra.lean"
|
|
|
|
|
|
rm -rf "$STASH"
|
|
|
|
|
|
}
|
|
|
|
|
|
trap cleanup EXIT INT TERM
|
|
|
|
|
|
|
|
|
|
|
|
cp "$HERE/Proofs/Audit.lean" "$STASH/Audit.lean"
|
|
|
|
|
|
cp "$HERE/AUDIT-MANIFEST.txt" "$STASH/AUDIT-MANIFEST.txt"
|
|
|
|
|
|
cp "$HERE/check.sh" "$STASH/check.sh"
|
|
|
|
|
|
|
|
|
|
|
|
DRIVER0B="$STASH/phase0b.sh"
|
|
|
|
|
|
{
|
|
|
|
|
|
echo 'set -uo pipefail'
|
|
|
|
|
|
echo "HERE=\"$HERE\""
|
|
|
|
|
|
sed -n '/^# ── Phase 0b/,/^# ── Phase 1/p' "$HERE/check.sh" | sed '$d'
|
|
|
|
|
|
} > "$DRIVER0B"
|
|
|
|
|
|
if [ "$(wc -l < "$DRIVER0B")" -lt 20 ]; then
|
|
|
|
|
|
echo "FATAL: could not lift Phase 0b out of check.sh."; exit 1
|
|
|
|
|
|
fi
|
|
|
|
|
|
|
|
|
|
|
|
DRIVER="$STASH/phase3c.sh"
|
|
|
|
|
|
build_driver() {
|
|
|
|
|
|
{
|
|
|
|
|
|
echo 'set -uo pipefail'
|
|
|
|
|
|
echo 'source ~/aeneas-toolchain/env.sh'
|
|
|
|
|
|
echo "HERE=\"$HERE\""
|
|
|
|
|
|
echo 'AENEAS_LEAN="$AENEAS_HOME/backends/lean"'
|
|
|
|
|
|
echo "TIMEOUT=$TIMEOUT; CORES=\"$CORES\""
|
|
|
|
|
|
# CERTS is referenced by the cross-check inside Phase 3c.
|
|
|
|
|
|
sed -n '/^CERTS=(/,/^)/p' "$HERE/check.sh"
|
|
|
|
|
|
# `$0` inside Phase 3c must resolve to the shipping check.sh, not to this
|
|
|
|
|
|
# driver, or the apex-name recovery would read the wrong file.
|
verification: pin the whole declaration surface (P1-b)
Phase 2b asks the kernel whether any AXIOM is declared under Proofs/. Phase 3
pins the cones of the named certificates. Between them sat every other
declaration in the corpus — around three thousand of them — and a helper lemma
quietly acquiring a hash oracle in its cone moved nothing either phase looked
at.
Phase 2c closes that. Ported from ltl-accumulator-verified, where a nine-attack
self-test proved a source-regex enumerator evadable by attributed, private,
indented and `instance` declarations and by a nested-namespace basename
collision. Reading the compiled environment sees what the kernel saw; no name
shape hides. Every constant contributes module, name, kind and full axiom cone,
and the observed set must equal inventory-allowlist.txt exactly in BOTH
directions, with a count trailer so a truncated run cannot pass as an empty
diff.
FOUR THINGS THIS BUILD GOT WRONG, each caught by a check rather than by review:
- The number of inventory drivers is a per-repo FACT, not an assumption.
dalek and anza cannot import their corpus as one environment (Proofs.Basic
and Proofs.ConstSpecs both declare CurveFieldProofs.zero_spec); risc0 and
betrusted have no Proofs.Basic at all. Determined by compiling a probe.
check.sh now DISCOVERS its drivers from the filesystem instead of naming
two, and the generator refuses to split out a module the repo lacks.
- The split let one real declaration hide behind another's entry. Keyed on
name alone, the two zero_specs produced byte-identical records, so 3022
declarations were covered by 3021 allowlist entries. Caught by the count
trailer. Every record now carries its originating module.
- The gate's success line said "single sanctioned axiom", inherited from the
accumulator's policy. This corpus permits NONE. A success message
describing a different rule is how an assertion stops meaning anything.
- selftest-axgate.sh lifted Phase 2b with a range ending at "Phase 3", so
inserting Phase 2c between them made it swallow the new phase and die on
variables only check.sh defines — surfacing as the BASELINE case failing,
a self-test blaming a gate for its own extraction bug. Both self-tests now
stop at the next phase marker whatever it is called, and refuse to run if
they capture more than one phase. The guard is the fix; the range was the
symptom.
WHAT THIS IS NOT, recorded in TRUSTED-BASE.md at the same length as the claim:
- No independent cone walker. The accumulator cross-checks collectAxioms
against a hand-written walker. Ported here it was wrong in BOTH directions
on mathlib's inductive shapes: EdPoint gave [] against the kernel's three
axioms, and once extended, ProjPoint gave three against the kernel's none.
Two implementations disagreeing both ways are a second wrong answer, not a
check. These cones rest on collectAxioms alone.
- Thirteen Proofs/Scalar* modules are inventoried by nothing — the
second-button seam, still open. Phase 2c names every uncovered module on
every run so the omission is visible rather than inferred.
selftest-inventory.sh exercises the shipping gate with six cases, each
asserting a specific diagnostic, including the one that matters: a cone
widened by one oracle while name, module and kind stay put. Negative-tested by
disabling the gate's diff, which turns two cases red including one for the
wrong reason, correctly reported as such.
Verified green: 20 runs across the four repositories (four buttons, four
harness, four inventory, four axgate, four binding self-tests), zero red. The
four check-scalar.sh greens from the preceding sweep stand: that script neither
reads the pin file nor changed.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-07-29 23:20:20 +00:00
|
|
|
|
# Stop at the next phase marker, not at a blank echo: a terminator that is
|
|
|
|
|
|
# not itself a phase boundary breaks the moment the phase's body changes.
|
verification: --audit-only mode, and the guard that keeps it from becoming evidence (T1)
Gate work dominates this estate's wall-clock: on 2026-07-29, 3.9 hours of a
session went to Lean re-elaborating proofs nobody had edited while the audit
phases themselves took about fifteen seconds. --audit-only runs every gate
against the artifacts a previous full run left behind: ~60s against ~1280s.
IT IS SAFE ONLY BECAUSE IT REFUSES.
- It requires every shipped .lean to be BYTE-IDENTICAL to a basis recorded by
a previous full run. Not mtimes: `touch` defeats those, and a stale-artifact
check that fails open is worse than no shortcut at all, because a green
button would then describe a corpus that is no longer on disk.
- The basis is gitignored build state, so a fresh clone cannot inherit
permission to skip compiling.
- The closing banner differs and says in words that the run is not evidence.
selftest-auditonly.sh exercises seven cases: no basis, an edited comment
character, a deleted source, a new source, a missing artifact, a truncated
basis, and — asserted as a PASS — every source's mtime touched with bytes
unchanged, which pins the bytes-not-mtimes decision rather than leaving it
implicit. Negative-tested: with the basis comparison disabled a changed source
is wrongly accepted, exit 0 and zero refusals, so the guard is load-bearing.
A PHASE TERMINATOR, because this broke twice. Every self-test lifts a phase
from check.sh by scanning to the next phase marker. The last phase had no
marker after it, so a lift ran to end-of-file and swallowed whatever was
appended later — first Phase 2c into the axgate lift, then T1's tail into the
binding lift, where it referenced $AUDIT_ONLY and died under `set -u`. Both
surfaced as the BASELINE case failing: a self-test blaming a gate for its own
extraction bug. The phases now end at an explicit sentinel and both lifters
stop there, so nothing appended below can silently become part of the last
phase from a lifter's point of view.
TRUSTED-BASE.md records what an audit-only transcript does and does not
establish, and — because it cost a confusing red run today — that lean-guard's
memory clamp presents as `FAIL: Proofs/<module>` while being a resource
condition, not a broken proof.
Verified green: 8 full button runs (four check.sh, four check-scalar.sh) and 20
self-tests across the four repositories, zero red. One earlier run failed on
the memory clamp because the author ran a test suite concurrently; re-run on a
quiet machine, green.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-07-30 17:16:21 +00:00
|
|
|
|
awk '/^# ── Phase 3c/{f=1} f&&/^# ── (Phase |Phases end)/&&!/Phase 3c/{exit} f{print}' "$HERE/check.sh" \
|
verification: bind the statements, the specifications, and the model (P1-a)
Phases 3/3b establish what each certificate RESTS ON. Neither says what it
SAYS, nor what it is ABOUT. A certificate gutted to a tautology of the same
axiom cone passes both; so does one whose reference definition has been
redefined to BE the extracted code, at which point the theorem reads
`loop = loop` and every cone is byte-identical.
Phase 3c closes that. Proofs/Audit.lean emits a canonical block holding the
policy constants, every certificate's fully-elaborated statement (pp.all, so
implicit arguments, instances and universe levels are visible), and the body
of every specification constant transitively reachable from those statements.
Its SHA-256 is pinned in check.sh and the block itself is committed as
AUDIT-MANIFEST.txt, so a mismatch is DIFFED, not merely reported. 31
certificates, 68 specification constants per repository.
Two tiers, not one. These forks have an arithmetic tier that must stay
oracle-free and an apex tier carrying this fork's hash and wire-format axioms,
and the apex boundary genuinely differs per fork (dalek 8 extra names, anza 4,
risc0 and betrusted 5). One shared constant would have widened the arithmetic
tier to accept hash oracles, which is the most valuable property these repos
have. Each auditor is generated from its own repository's policy.
Phase 0b pins the extracted model. This was not a precaution: risc0 and
betrusted were observed emitting BYTE-IDENTICAL audit-manifest digests
(6c821b8e…) while shipping demonstrably different extracted models, their
point-doubling routines differing in operation order. A statement names an
extracted function; it does not contain that function's body. Binding
statements is not binding the subject. Membership derives from the filesystem,
so a new model file fails closed.
selftest-statements.sh attacks both phases with ten cases, each asserting a
specific diagnostic: an edited model body, an unlisted model file, a widened
policy, a hand-edited committed block, a certificate dropped from the auditor
WITH the digest refreshed to match, and a gutted statement whose cone is
unchanged. It lifts the phases out of check.sh at run time, so it attacks the
shipping gate rather than a copy.
Two bugs found and fixed during that testing, both mine: Phase 3c read `$0`
after `cd "$AENEAS_LEAN"`, and $0 is the caller's relative path; and the
axgate self-test compared the tree against a pristine checkout rather than
against how it found it. A third expectation was wrong rather than the code —
widening the apex boundary is caught by the exact-cone requirement before the
digest ever runs, which is a stronger rejection, and the test now says so.
All sixteen runs green at these commits: four main buttons, four axgate
self-tests, four binding self-tests, four scalar buttons.
TRUSTED-BASE.md records what this binds and, at equal length, what it does
not: a digest binds identity, not meaning; an author can rotate the pins in
one commit and is caught by review, not by the script; and pinning the model
says nothing about whether Charon and Aeneas translated the Rust faithfully.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-07-28 22:38:22 +00:00
|
|
|
|
| sed "s|\"\$0\"|\"$HERE/check.sh\"|g"
|
|
|
|
|
|
} > "$DRIVER"
|
|
|
|
|
|
if [ "$(wc -l < "$DRIVER")" -lt 60 ]; then
|
|
|
|
|
|
echo "FATAL: could not lift Phase 3c out of check.sh — the phase markers moved."
|
|
|
|
|
|
echo "This self-test must attack the shipping gate; refusing to run against nothing."
|
|
|
|
|
|
exit 1
|
|
|
|
|
|
fi
|
|
|
|
|
|
}
|
|
|
|
|
|
build_driver
|
|
|
|
|
|
|
|
|
|
|
|
recompile_audit() {
|
|
|
|
|
|
( cd "$AENEAS_LEAN" && lake env bash -c "
|
|
|
|
|
|
set -uo pipefail
|
|
|
|
|
|
cd '$HERE/gen' && export LEAN_PATH=\"\$LEAN_PATH:\$PWD:$HERE\"
|
|
|
|
|
|
cd '$HERE'
|
|
|
|
|
|
LEAN_TIMEOUT=$TIMEOUT '$HERE/lean-guard' Proofs/Audit.lean
|
|
|
|
|
|
" ) >/dev/null 2>&1
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
expect() { # expect <name> <expected-rc> <required-substring>
|
|
|
|
|
|
local name="$1" want_rc="$2" want_txt="$3" out rc
|
|
|
|
|
|
out=$(bash "$DRIVER" 2>&1); rc=$?
|
|
|
|
|
|
if [ "$rc" -ne "$want_rc" ]; then
|
|
|
|
|
|
echo " FAIL $name: exit $rc, expected $want_rc"; FAILURES=$((FAILURES+1)); return
|
|
|
|
|
|
fi
|
|
|
|
|
|
if ! grep -qF "$want_txt" <<<"$out"; then
|
|
|
|
|
|
echo " FAIL $name: exit code right, diagnostic wrong (rejected for the wrong reason)"
|
|
|
|
|
|
echo " wanted: $want_txt"
|
|
|
|
|
|
echo " got: $(tr '\n' '|' <<<"$out" | cut -c1-260)"
|
|
|
|
|
|
FAILURES=$((FAILURES+1)); return
|
|
|
|
|
|
fi
|
|
|
|
|
|
echo " ok $name"
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
echo "=== selftest-statements: attacking check.sh Phases 0b and 3c ==="
|
|
|
|
|
|
|
|
|
|
|
|
# ── 0. THE SUBJECT. The certificates are stated ABOUT the extracted model in
|
|
|
|
|
|
# gen/. A statement names an extracted function; editing that function's
|
|
|
|
|
|
# BODY changes what the theorem is about while leaving every statement,
|
|
|
|
|
|
# every cone and the Phase 3c digest byte-identical. Demonstrated on
|
|
|
|
|
|
# 2026-07-28: risc0 and betrusted ship different extracted models and
|
|
|
|
|
|
# produce the SAME audit-manifest digest. Only the byte pin separates them.
|
|
|
|
|
|
expect0b() {
|
|
|
|
|
|
local name="$1" want_rc="$2" want_txt="$3" out rc
|
|
|
|
|
|
out=$(bash "$DRIVER0B" 2>&1); rc=$?
|
|
|
|
|
|
if [ "$rc" -ne "$want_rc" ]; then
|
|
|
|
|
|
echo " FAIL $name: exit $rc, expected $want_rc"; FAILURES=$((FAILURES+1)); return
|
|
|
|
|
|
fi
|
|
|
|
|
|
if ! grep -qF "$want_txt" <<<"$out"; then
|
|
|
|
|
|
echo " FAIL $name: exit code right, diagnostic wrong (rejected for the wrong reason)"
|
|
|
|
|
|
echo " wanted: $want_txt"; FAILURES=$((FAILURES+1)); return
|
|
|
|
|
|
fi
|
|
|
|
|
|
echo " ok $name"
|
|
|
|
|
|
}
|
|
|
|
|
|
expect0b "model pin: baseline green" 0 "match their pins"
|
|
|
|
|
|
GENF=$(awk 'NR==1{print $2}' "$HERE/GEN-MODEL.sha256")
|
|
|
|
|
|
cp "$HERE/gen/$GENF" "$STASH/genfile.bak"
|
|
|
|
|
|
printf '\n-- edited\n' >> "$HERE/gen/$GENF"
|
|
|
|
|
|
expect0b "model pin: edited model body caught" 1 "does not match its pin"
|
|
|
|
|
|
cp "$STASH/genfile.bak" "$HERE/gen/$GENF"
|
|
|
|
|
|
cp "$HERE/gen/$GENF" "$HERE/gen/$(dirname "$GENF")/ZZExtra.lean"
|
|
|
|
|
|
expect0b "model pin: an unlisted model file caught" 1 "does not match GEN-MODEL.sha256"
|
|
|
|
|
|
rm -f "$HERE/gen/$(dirname "$GENF")/ZZExtra.lean"
|
|
|
|
|
|
|
|
|
|
|
|
# ── 1. Baseline: untouched repository passes and reports what it bound.
|
|
|
|
|
|
expect "baseline green, manifest digest matches" 0 "audit-manifest sha256"
|
|
|
|
|
|
|
|
|
|
|
|
# ── 2. WIDEN THE POLICY. Add one name to the apex boundary.
|
|
|
|
|
|
# The policy IS inside the digest, but this attack never reaches the
|
|
|
|
|
|
# digest: because the apex tier requires EXACT cone equality, a widened
|
|
|
|
|
|
# boundary immediately shows up as `missing=` on all four apex
|
|
|
|
|
|
# certificates. That is a strictly stronger rejection than a digest
|
|
|
|
|
|
# mismatch — it names which certificates stopped matching instead of
|
|
|
|
|
|
# merely reporting that some byte moved. Asserted here as such, because a
|
|
|
|
|
|
# test that demanded the weaker diagnostic would go red the day the
|
|
|
|
|
|
# stronger guard started working.
|
|
|
|
|
|
python3 - "$HERE/Proofs/Audit.lean" <<'PY'
|
|
|
|
|
|
import sys, re
|
|
|
|
|
|
f = sys.argv[1]; s = open(f).read()
|
|
|
|
|
|
m = re.search(r'^def apexExtra : List Name :=\n \[(.*)\]$', s, re.M)
|
|
|
|
|
|
assert m, "apexExtra not found"
|
|
|
|
|
|
s = s.replace(m.group(0), m.group(0)[:-1] + ", `Classical.byContradiction]", 1)
|
|
|
|
|
|
open(f, "w").write(s)
|
|
|
|
|
|
PY
|
|
|
|
|
|
recompile_audit
|
|
|
|
|
|
expect "widened policy caught by the exact-cone requirement" 1 "missing=[Classical.byContradiction]"
|
|
|
|
|
|
cp "$STASH/Audit.lean" "$HERE/Proofs/Audit.lean"; recompile_audit
|
|
|
|
|
|
|
|
|
|
|
|
# ── 3. HAND-EDIT THE COMMITTED BLOCK. The digest still matches the emitted
|
|
|
|
|
|
# block, so only the byte-comparison against the committed input can see
|
|
|
|
|
|
# this. Without it the diff printed on a future failure would silently
|
|
|
|
|
|
# compare against a doctored reference.
|
|
|
|
|
|
sed -i '2s/$/ TAMPERED/' "$HERE/AUDIT-MANIFEST.txt"
|
|
|
|
|
|
expect "hand-edited committed block caught" 1 "does not match the emitted block"
|
|
|
|
|
|
cp "$STASH/AUDIT-MANIFEST.txt" "$HERE/AUDIT-MANIFEST.txt"
|
|
|
|
|
|
|
|
|
|
|
|
# ── 4. DROP A CERTIFICATE FROM THE AUDITOR — AND REFRESH THE DIGEST TO MATCH,
|
|
|
|
|
|
# exactly as an author covering their tracks would. The digest and the
|
|
|
|
|
|
# committed block are now perfectly consistent with each other. The only
|
|
|
|
|
|
# thing that can still object is the cross-check against the certificate
|
|
|
|
|
|
# list this script derives from its OWN two sources.
|
|
|
|
|
|
python3 - "$HERE/Proofs/Audit.lean" <<'PY'
|
|
|
|
|
|
import sys, re
|
|
|
|
|
|
f = sys.argv[1]; s = open(f).read()
|
|
|
|
|
|
lines = s.splitlines(True)
|
|
|
|
|
|
i = next(k for k, l in enumerate(lines) if l.strip().startswith(', (`') and 'kernel3)' in l)
|
|
|
|
|
|
del lines[i]
|
|
|
|
|
|
open(f, "w").write("".join(lines))
|
|
|
|
|
|
PY
|
|
|
|
|
|
recompile_audit
|
|
|
|
|
|
# Regenerate the committed block and re-pin the digest from the tampered run.
|
|
|
|
|
|
TAMPERED_OUT=$(cd "$AENEAS_LEAN" && lake env bash -c "
|
|
|
|
|
|
set -uo pipefail
|
|
|
|
|
|
cd '$HERE/gen' && export LEAN_PATH=\"\$LEAN_PATH:\$PWD:$HERE\"
|
|
|
|
|
|
cd '$HERE'
|
|
|
|
|
|
LEAN_TIMEOUT=$TIMEOUT '$HERE/lean-guard' Proofs/Audit.lean 2>&1")
|
|
|
|
|
|
awk '/AUDIT-MANIFEST-BEGIN/{f=1;next} /AUDIT-MANIFEST-END/{f=0} f' <<<"$TAMPERED_OUT" > "$HERE/AUDIT-MANIFEST.txt"
|
|
|
|
|
|
NEWSHA=$(sha256sum "$HERE/AUDIT-MANIFEST.txt" | cut -d' ' -f1)
|
|
|
|
|
|
sed -i "s/^EXPECTED_AUDIT_SHA256=.*/EXPECTED_AUDIT_SHA256=\"$NEWSHA\"/" "$HERE/check.sh"
|
|
|
|
|
|
build_driver
|
|
|
|
|
|
expect "dropped certificate caught by the cross-check, digest refreshed or not" 1 "have drifted"
|
|
|
|
|
|
cp "$STASH/Audit.lean" "$HERE/Proofs/Audit.lean"
|
|
|
|
|
|
cp "$STASH/AUDIT-MANIFEST.txt" "$HERE/AUDIT-MANIFEST.txt"
|
|
|
|
|
|
cp "$STASH/check.sh" "$HERE/check.sh"
|
|
|
|
|
|
build_driver
|
|
|
|
|
|
recompile_audit
|
|
|
|
|
|
|
|
|
|
|
|
# ── 5. THE REAL ONE: gut a certificate's STATEMENT while preserving its axiom
|
|
|
|
|
|
# cone. Every earlier phase passes this; only the statement binding sees it.
|
|
|
|
|
|
if [ "$SKIP_SLOW" = "1" ]; then
|
|
|
|
|
|
echo " skip gutted-statement case (SKIP_SLOW=1) — the fast gates above do not cover it"
|
|
|
|
|
|
else
|
|
|
|
|
|
# A TERMINAL certificate: nothing else in the corpus consumes it. Gutting a
|
|
|
|
|
|
# load-bearing one simply breaks its consumers and the module stops
|
|
|
|
|
|
# compiling, which demonstrates the compiler working, not this gate.
|
|
|
|
|
|
GUT_MOD=FieldMain
|
|
|
|
|
|
GUT_CERT=CurveFieldProofs.fieldImplementation
|
|
|
|
|
|
cp "$HERE/Proofs/$GUT_MOD.lean" "$STASH/gut.lean"
|
|
|
|
|
|
python3 - "$HERE/Proofs/$GUT_MOD.lean" "$GUT_CERT" <<'PY'
|
|
|
|
|
|
import sys, re
|
|
|
|
|
|
f, cert = sys.argv[1], sys.argv[2]
|
|
|
|
|
|
short = cert.split('.')[-1]
|
|
|
|
|
|
s = open(f).read()
|
|
|
|
|
|
m = re.search(r'^theorem %s\b' % re.escape(short), s, re.M)
|
|
|
|
|
|
assert m, f"certificate {short} not found in {f}"
|
|
|
|
|
|
i = m.start()
|
|
|
|
|
|
# find the next top-level declaration after it
|
|
|
|
|
|
nxt = re.search(r'^(theorem|lemma|def|noncomputable def|end|/--|@\[)', s[i+10:], re.M)
|
|
|
|
|
|
assert nxt, "no following declaration"
|
|
|
|
|
|
j = i + 10 + nxt.start()
|
|
|
|
|
|
# Same cone (Classical.em pulls in Classical.choice/propext), utterly different claim.
|
|
|
|
|
|
gut = "theorem %s : (∀ p : Prop, p ∨ ¬p) := Classical.em\n\n" % short
|
|
|
|
|
|
open(f, "w").write(s[:i] + gut + s[j:])
|
|
|
|
|
|
PY
|
|
|
|
|
|
( cd "$AENEAS_LEAN" && lake env bash -c "
|
|
|
|
|
|
set -uo pipefail
|
|
|
|
|
|
cd '$HERE/gen' && export LEAN_PATH=\"\$LEAN_PATH:\$PWD:$HERE\"
|
|
|
|
|
|
cd '$HERE'
|
|
|
|
|
|
LEAN_TIMEOUT=$TIMEOUT '$HERE/lean-guard' Proofs/$GUT_MOD.lean
|
|
|
|
|
|
" ) >/dev/null 2>&1 || { echo " FAIL setup: the gutted module did not compile"; FAILURES=$((FAILURES+1)); }
|
|
|
|
|
|
recompile_audit
|
|
|
|
|
|
expect "gutted statement caught (cone unchanged)" 1 "audit-manifest digest mismatch"
|
|
|
|
|
|
cp "$STASH/gut.lean" "$HERE/Proofs/$GUT_MOD.lean"
|
|
|
|
|
|
( cd "$AENEAS_LEAN" && lake env bash -c "
|
|
|
|
|
|
set -uo pipefail
|
|
|
|
|
|
cd '$HERE/gen' && export LEAN_PATH=\"\$LEAN_PATH:\$PWD:$HERE\"
|
|
|
|
|
|
cd '$HERE'
|
|
|
|
|
|
LEAN_TIMEOUT=$TIMEOUT '$HERE/lean-guard' Proofs/$GUT_MOD.lean
|
|
|
|
|
|
" ) >/dev/null 2>&1
|
|
|
|
|
|
recompile_audit
|
|
|
|
|
|
fi
|
|
|
|
|
|
|
|
|
|
|
|
# ── 6. Restored: green again, and the working tree is as we found it.
|
|
|
|
|
|
expect "restored to green" 0 "audit-manifest sha256"
|
verification: pin the harness, the audit drivers and the policy files (P1-c)
Every gate this repository has was executed by scripts that nothing pinned.
Round-5 review of the companion SLH-DSA repository stubbed the compiler
wrapper alone and its button printed ALL GREEN in 3.6 seconds over
deliberately destroyed proofs; flipping two guards in the audit driver
disabled every check with the digest byte-identical. Depth of checking is
worth nothing if the thing doing the checking is unbound — and every gate
added this week made that gap more valuable to an attacker, not less.
Phase 0c requires every harness file to match HARNESS.sha256. Two design
points carry the weight:
- WHICH files must be pinned is POLICY and lives in check.sh, never in the
map being consulted. If the required set were read from the pin file,
deleting an entry would silently un-pin that file. It is instead derived
from the filesystem, so a deleted entry is a set mismatch and a build
failure. That is the exact defect SLH-DSA round-6 found, closed here by
construction.
- Membership self-derives from the executable bit: anything this script can
shell out to must be pinned, so a NEW script fails closed until someone
pins it deliberately. Load-bearing files that are not executable — the
audit driver, the committed manifests, the policy tables — cannot be
discovered that way and are listed explicitly.
lean-guard is inside the set, which finally makes the standing "lean-guard
stays hash-pinned" rule a property of the repository rather than a convention.
selftest-harness.sh replays five cases, each asserting a specific diagnostic:
an edited lean-guard, a new unpinned executable, a deleted pin entry, a
missing pin file, and a positive control. It was itself negative-tested — with
the hash comparison removed it goes red on exactly that case while cheerfully
reporting "10 harness files match their pins".
TRUSTED-BASE.md states the limit at equal length to the claim: pinning a
harness from inside that harness is circular, and an author who edits a script
and refreshes its pin in the same commit passes every phase. What the pin
changes is that the edit can no longer be SILENT — it must appear in the diff
at the commit being reviewed. A green button says "this is the apparatus that
was reviewed", never "this apparatus is trustworthy".
Also fixed, found by this sweep: both self-tests compared the working tree
against its starting state with `diff <(echo "$VAR") <(command)`, which is
asymmetric — for a clean tree the variable is empty and `echo` emits a blank
line the command does not. It reported a difference precisely when nothing was
wrong, and only surfaced once P1-a was committed and Proofs/ became clean.
Both now compare as strings.
Verified green: 20 runs across the four ed25519 repositories (four buttons,
four harness self-tests, four axiom-gate self-tests, four binding self-tests,
four scalar buttons), zero red.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-07-29 18:13:01 +00:00
|
|
|
|
# String comparison, not diff of process substitutions: for a clean tree the
|
|
|
|
|
|
# variable is empty and `echo` emits a blank line the raw command does not,
|
|
|
|
|
|
# which reports a difference precisely when there is none.
|
|
|
|
|
|
TREE_NOW="$(cd "$HERE/.." && git status --porcelain)"
|
|
|
|
|
|
if [ "$TREE_AT_START" != "$TREE_NOW" ]; then
|
verification: bind the statements, the specifications, and the model (P1-a)
Phases 3/3b establish what each certificate RESTS ON. Neither says what it
SAYS, nor what it is ABOUT. A certificate gutted to a tautology of the same
axiom cone passes both; so does one whose reference definition has been
redefined to BE the extracted code, at which point the theorem reads
`loop = loop` and every cone is byte-identical.
Phase 3c closes that. Proofs/Audit.lean emits a canonical block holding the
policy constants, every certificate's fully-elaborated statement (pp.all, so
implicit arguments, instances and universe levels are visible), and the body
of every specification constant transitively reachable from those statements.
Its SHA-256 is pinned in check.sh and the block itself is committed as
AUDIT-MANIFEST.txt, so a mismatch is DIFFED, not merely reported. 31
certificates, 68 specification constants per repository.
Two tiers, not one. These forks have an arithmetic tier that must stay
oracle-free and an apex tier carrying this fork's hash and wire-format axioms,
and the apex boundary genuinely differs per fork (dalek 8 extra names, anza 4,
risc0 and betrusted 5). One shared constant would have widened the arithmetic
tier to accept hash oracles, which is the most valuable property these repos
have. Each auditor is generated from its own repository's policy.
Phase 0b pins the extracted model. This was not a precaution: risc0 and
betrusted were observed emitting BYTE-IDENTICAL audit-manifest digests
(6c821b8e…) while shipping demonstrably different extracted models, their
point-doubling routines differing in operation order. A statement names an
extracted function; it does not contain that function's body. Binding
statements is not binding the subject. Membership derives from the filesystem,
so a new model file fails closed.
selftest-statements.sh attacks both phases with ten cases, each asserting a
specific diagnostic: an edited model body, an unlisted model file, a widened
policy, a hand-edited committed block, a certificate dropped from the auditor
WITH the digest refreshed to match, and a gutted statement whose cone is
unchanged. It lifts the phases out of check.sh at run time, so it attacks the
shipping gate rather than a copy.
Two bugs found and fixed during that testing, both mine: Phase 3c read `$0`
after `cd "$AENEAS_LEAN"`, and $0 is the caller's relative path; and the
axgate self-test compared the tree against a pristine checkout rather than
against how it found it. A third expectation was wrong rather than the code —
widening the apex boundary is caught by the exact-cone requirement before the
digest ever runs, which is a stronger rejection, and the test now says so.
All sixteen runs green at these commits: four main buttons, four axgate
self-tests, four binding self-tests, four scalar buttons.
TRUSTED-BASE.md records what this binds and, at equal length, what it does
not: a digest binds identity, not meaning; an author can rotate the pins in
one commit and is caught by review, not by the script; and pinning the model
says nothing about whether Charon and Aeneas translated the Rust faithfully.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-07-28 22:38:22 +00:00
|
|
|
|
echo " FAIL restore: the working tree differs from how this test found it:"
|
verification: pin the harness, the audit drivers and the policy files (P1-c)
Every gate this repository has was executed by scripts that nothing pinned.
Round-5 review of the companion SLH-DSA repository stubbed the compiler
wrapper alone and its button printed ALL GREEN in 3.6 seconds over
deliberately destroyed proofs; flipping two guards in the audit driver
disabled every check with the digest byte-identical. Depth of checking is
worth nothing if the thing doing the checking is unbound — and every gate
added this week made that gap more valuable to an attacker, not less.
Phase 0c requires every harness file to match HARNESS.sha256. Two design
points carry the weight:
- WHICH files must be pinned is POLICY and lives in check.sh, never in the
map being consulted. If the required set were read from the pin file,
deleting an entry would silently un-pin that file. It is instead derived
from the filesystem, so a deleted entry is a set mismatch and a build
failure. That is the exact defect SLH-DSA round-6 found, closed here by
construction.
- Membership self-derives from the executable bit: anything this script can
shell out to must be pinned, so a NEW script fails closed until someone
pins it deliberately. Load-bearing files that are not executable — the
audit driver, the committed manifests, the policy tables — cannot be
discovered that way and are listed explicitly.
lean-guard is inside the set, which finally makes the standing "lean-guard
stays hash-pinned" rule a property of the repository rather than a convention.
selftest-harness.sh replays five cases, each asserting a specific diagnostic:
an edited lean-guard, a new unpinned executable, a deleted pin entry, a
missing pin file, and a positive control. It was itself negative-tested — with
the hash comparison removed it goes red on exactly that case while cheerfully
reporting "10 harness files match their pins".
TRUSTED-BASE.md states the limit at equal length to the claim: pinning a
harness from inside that harness is circular, and an author who edits a script
and refreshes its pin in the same commit passes every phase. What the pin
changes is that the edit can no longer be SILENT — it must appear in the diff
at the commit being reviewed. A green button says "this is the apparatus that
was reviewed", never "this apparatus is trustworthy".
Also fixed, found by this sweep: both self-tests compared the working tree
against its starting state with `diff <(echo "$VAR") <(command)`, which is
asymmetric — for a clean tree the variable is empty and `echo` emits a blank
line the command does not. It reported a difference precisely when nothing was
wrong, and only surfaced once P1-a was committed and Proofs/ became clean.
Both now compare as strings.
Verified green: 20 runs across the four ed25519 repositories (four buttons,
four harness self-tests, four axiom-gate self-tests, four binding self-tests,
four scalar buttons), zero red.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-07-29 18:13:01 +00:00
|
|
|
|
diff <(printf '%s\n' "$TREE_AT_START") <(printf '%s\n' "$TREE_NOW") | sed 's/^/ /'
|
verification: bind the statements, the specifications, and the model (P1-a)
Phases 3/3b establish what each certificate RESTS ON. Neither says what it
SAYS, nor what it is ABOUT. A certificate gutted to a tautology of the same
axiom cone passes both; so does one whose reference definition has been
redefined to BE the extracted code, at which point the theorem reads
`loop = loop` and every cone is byte-identical.
Phase 3c closes that. Proofs/Audit.lean emits a canonical block holding the
policy constants, every certificate's fully-elaborated statement (pp.all, so
implicit arguments, instances and universe levels are visible), and the body
of every specification constant transitively reachable from those statements.
Its SHA-256 is pinned in check.sh and the block itself is committed as
AUDIT-MANIFEST.txt, so a mismatch is DIFFED, not merely reported. 31
certificates, 68 specification constants per repository.
Two tiers, not one. These forks have an arithmetic tier that must stay
oracle-free and an apex tier carrying this fork's hash and wire-format axioms,
and the apex boundary genuinely differs per fork (dalek 8 extra names, anza 4,
risc0 and betrusted 5). One shared constant would have widened the arithmetic
tier to accept hash oracles, which is the most valuable property these repos
have. Each auditor is generated from its own repository's policy.
Phase 0b pins the extracted model. This was not a precaution: risc0 and
betrusted were observed emitting BYTE-IDENTICAL audit-manifest digests
(6c821b8e…) while shipping demonstrably different extracted models, their
point-doubling routines differing in operation order. A statement names an
extracted function; it does not contain that function's body. Binding
statements is not binding the subject. Membership derives from the filesystem,
so a new model file fails closed.
selftest-statements.sh attacks both phases with ten cases, each asserting a
specific diagnostic: an edited model body, an unlisted model file, a widened
policy, a hand-edited committed block, a certificate dropped from the auditor
WITH the digest refreshed to match, and a gutted statement whose cone is
unchanged. It lifts the phases out of check.sh at run time, so it attacks the
shipping gate rather than a copy.
Two bugs found and fixed during that testing, both mine: Phase 3c read `$0`
after `cd "$AENEAS_LEAN"`, and $0 is the caller's relative path; and the
axgate self-test compared the tree against a pristine checkout rather than
against how it found it. A third expectation was wrong rather than the code —
widening the apex boundary is caught by the exact-cone requirement before the
digest ever runs, which is a stronger rejection, and the test now says so.
All sixteen runs green at these commits: four main buttons, four axgate
self-tests, four binding self-tests, four scalar buttons.
TRUSTED-BASE.md records what this binds and, at equal length, what it does
not: a digest binds identity, not meaning; an author can rotate the pins in
one commit and is caught by review, not by the script; and pinning the model
says nothing about whether Charon and Aeneas translated the Rust faithfully.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-07-28 22:38:22 +00:00
|
|
|
|
FAILURES=$((FAILURES+1))
|
|
|
|
|
|
else
|
|
|
|
|
|
echo " ok working tree restored to its starting state"
|
|
|
|
|
|
fi
|
|
|
|
|
|
|
|
|
|
|
|
echo ""
|
|
|
|
|
|
if [ "$FAILURES" -eq 0 ]; then
|
|
|
|
|
|
echo "SELFTEST PASSED — Phase 3c rejects statement- and specification-level"
|
|
|
|
|
|
echo "tampering that moves no axiom cone, for the stated reason in each case."
|
|
|
|
|
|
exit 0
|
|
|
|
|
|
fi
|
|
|
|
|
|
echo "SELFTEST FAILED: $FAILURES check(s) did not behave as claimed."
|
|
|
|
|
|
exit 1
|