AERE-CROSS-CLIENT-DETERMINISM.md

@@ -165,8 +165,7 @@ hash and state root are compared to `rpc.aere.network`.
 
 **Evidence.**
 - `aerenew/docs/NETHERMIND-2ND-CLIENT-LIVE-2026-07-17.md`: the permanent second client
-  `aere-client2` (a dedicated second-client host, Helsinki, dedicated firewall,
-  JSON-RPC bound to `127.0.0.1` only, durable `systemd` service). Parity table of hash and
+  (a dedicated host, not publicly reachable, run as a durable service). Parity table of hash and
   state root MATCH at 1 / 100 / 1,000 / 10,000 / 11,133 / 11,136; zero invalid-seal rejects
   in range.
 - The first live-devp2p run (recorded in the memory note `client_diversity_2026-07-13` and

AERE-ENGINEERING-SECURITY-SPEC.md

@@ -269,7 +269,7 @@ A safety-first summary that holds across every mode tested: **in all adversarial
 
 **Halt / fork / continue.** The minority side halts; the majority side (if one partition holds >= 5 validators) continues; on reconnection the halted side recovers. No fork.
 
-**Status.** PROVEN (halt-then-recover demonstrated). Evidence: sub-quorum halt and in-place recovery were shown in the in-place PQC activation work (dropping below the required quorum halted block production at a specific height, and restoring the missing signers resumed from that exact height with no re-genesis) and in the hybrid-consensus negative controls (killing enough validators to break quorum halted the chain with no split and recovered in place). The single-operator caveat applies and is disclosed: all seven validators are Foundation-operated on Hetzner infrastructure, so a partition that is really an infrastructure or operator outage is a correlated failure, not an independent one. QBFT's partition safety (no fork) holds regardless; the availability of a quorum-holding side is an operational, not a protocol, guarantee at N=7 under one operator.
+**Status.** PROVEN (halt-then-recover demonstrated). Evidence: sub-quorum halt and in-place recovery were shown in the in-place PQC activation work (dropping below the required quorum halted block production at a specific height, and restoring the missing signers resumed from that exact height with no re-genesis) and in the hybrid-consensus negative controls (killing enough validators to break quorum halted the chain with no split and recovered in place). The single-operator caveat applies and is disclosed: all seven validators are Foundation-operated on a single cloud provider, so a partition that is really an infrastructure or operator outage is a correlated failure, not an independent one. QBFT's partition safety (no fork) holds regardless; the availability of a quorum-holding side is an operational, not a protocol, guarantee at N=7 under one operator.
 
 ### 5.5 The two clients disagree on state (Nethermind vs Besu)
 

AERE-FAULT-INJECTION-HARNESS.md

@@ -61,7 +61,7 @@ Under `aerenew/chaos/`:
   process-mode fallback that reuses the existing `audit-package-pq-consensus/harness/n7/` scripts.
 - `run-chaos-matrix.sh` — reads the matrix, injects each fault, snapshots per-node heights across
   the fault + recovery windows, decides halt/resume/fork, and (on a halt-then-recover) invokes the
-  attester. Has a hard guard that REFUSES to run on the live infra box.
+  attester. Has a hard guard that REFUSES to run on a production host.
 - `lib/partition.sh` — network partition via pumba netem / iptables so the majority keeps quorum.
 - `lib/attest-recovery.js` — ethers script that reads the record hash from the deployed registry,
   signs it with the operator's offline Falcon-512 key, and submits `submitRecovery(...)`.

AERE-NIST-VALIDATION-STATUS.md

@@ -55,7 +55,7 @@ For the finalized signature schemes (ML-DSA-44 and SLH-DSA-128s) the reject vect
 Validation is performed at more than one layer, and all layers must agree:
 
 - Solidity / EVM layer. Hardhat test suites deploy the on-chain verifier contracts and drive them against the committed JSON fixtures. Examples: `aerenew/contracts/test/mldsa44Verifier.test.js`, `aerenew/contracts/test/AereSphincsVerifier.test.js`, `aerenew/contracts/test/AereFalcon512Verifier.test.js`, `aerenew/contracts/test/AereFalcon1024Verifier.test.js`.
-- Native precompile layer. The precompile implementations themselves (Bouncy Castle backed) are exercised by a Java KAT harness on the build box, driven by `aerenew/parallel/combined-fork-binary/scripts/run_kat.sh`, which compiles and runs the probe against `bcprov-jdk18on-1.83.jar`. The committed result of that run is `aerenew/parallel/combined-fork-binary/logs/nist_kat_bc183.txt`: Falcon-512 and Falcon-1024 valid plus reject, ML-DSA-44 15/15, SLH-DSA-128s 14/14, SHAKE256 4/4, zero mismatches.
+- Native precompile layer. The precompile implementations themselves (Bouncy Castle backed) are exercised by a Java KAT harness on a separate build host, driven by `aerenew/parallel/combined-fork-binary/scripts/run_kat.sh`, which compiles and runs the probe against `bcprov-jdk18on-1.83.jar`. The committed result of that run is `aerenew/parallel/combined-fork-binary/logs/nist_kat_bc183.txt`: Falcon-512 and Falcon-1024 valid plus reject, ML-DSA-44 15/15, SLH-DSA-128s 14/14, SHAKE256 4/4, zero mismatches.
 - Live on-chain layer. After the fork cross, all five mainnet precompiles were exercised by direct `eth_call` against a running node (`aerenew/parallel/combined-fork-binary/scripts/check5_precompiles.py`), and a verify-and-record probe wrote accept/reject results into contract state (`aerenew/dryrun-results/dryrun/phase_postfork_kat.sh`), confirming that an invalid vector reverts on-chain and a valid one records success.
 
 ### 3.3 Underlying implementation
@@ -84,7 +84,7 @@ The point of self-validation is that anyone can rerun it. The committed inputs (
 
 ### Appendix A: items flagged for human verification before publishing
 
-- `[VERIFY]` The native Bouncy Castle KAT harness source (referred to as `Probe.java` in `run_kat.sh`) and the flat vector files it consumes (`falcon512.txt`, `falcon1024.txt`, `mldsa44.txt`, `slhdsa128s.txt`, `shake256.txt`, expected on the build box at `/root/work/probe/` and `/opt/aere/besu-pqc-fork/probe/`) are NOT committed to this repository. What IS committed is the result log (`nist_kat_bc183.txt`), the JSON fixtures, the Solidity/JS harnesses, and the run scripts. Decide whether to commit `Probe.java` and the flat vector files so the native-layer run is fully reproducible from the repo alone.
+- `[VERIFY]` The native Bouncy Castle KAT harness source (referred to as `Probe.java` in `run_kat.sh`) and the flat vector files it consumes (`falcon512.txt`, `falcon1024.txt`, `mldsa44.txt`, `slhdsa128s.txt`, `shake256.txt`, expected on a separate build host at `<work>/probe/` and `<fork>/probe/`) are NOT committed to this repository. What IS committed is the result log (`nist_kat_bc183.txt`), the JSON fixtures, the Solidity/JS harnesses, and the run scripts. Decide whether to commit `Probe.java` and the flat vector files so the native-layer run is fully reproducible from the repo alone.
 - `[VERIFY]` SHAKE256 coverage is described as KAT (a small set of FIPS 202 known-answer digests), not full ACVP SHA-3/SHAKE suite coverage. If a stronger claim is ever wanted, commit and run the complete NIST ACVP SHA-3 vector set first.
 - `[VERIFY]` The Falcon KAT provenance cited in the harness comment is `falcon512-KAT.rsp`, vector 0, SHA-256 `dd75c946...b5cd`, reproduced from the round-3 submission. Only the extracted vector-0 (and vector-1) JSON is committed, not the original `.rsp` file. Confirm the `.rsp` source hash independently if provenance is challenged.
 - `[VERIFY]` Confirm the block number for mainnet precompile activation (9,189,161) and the precompile addresses `0x0AE1` through `0x0AE5` against the live chain before publishing, and confirm ML-KEM-768 (`0x0AE6`) and Falcon HashToPoint (`0x0AE7`) remain testnet only.

AERE-REPRODUCIBLE-BUILDS.md

@@ -472,7 +472,7 @@ bash setup-fork.sh     # clones besu@d2032017, overlays fork EVM, adds 0x0AE6/0x
 bash build-dist.sh     # ./gradlew :besu:installDist -x test (one full build)
 ```
 
-[MEASURE: heavy Linux-only build, targets `/root/besu` on a scratch box. Not run here, and never on
+[MEASURE: heavy Linux-only build, targets `$BESU_HOME` on a scratch box. Not run here, and never on
 the live infra box, which runs production services. Publish the resulting `bin/besu` and jar SHA256
 values and compare against `combined-fork-binary/BUILD_INFO.txt`.]
 

REPRODUCE.md

@@ -352,8 +352,8 @@ Committed results, to be reconfirmed by a fresh run:
   against an independent Python `hashlib.shake_256` oracle
   [CITED: aerenew/pqc-fork/results/kat-results-hashtopoint.json].
 
-[MEASURE: `run-kats.sh` targets `/root/besu` and `/root/staging` on the isolated fork host and
-cannot run on this machine or on the live infra box (heavy build; the infra box runs live services
+[MEASURE: `run-kats.sh` targets `$BESU_HOME` and `$STAGING` on an isolated build host and
+cannot run on this machine or on any production host (heavy build; the infra box runs live services
 and must never take heavy compute). Reproduce on a scratch Linux box and publish the two result
 JSONs. The Rust acceleration crate has its own KATs: `cd aerenew/pqc-accel && cargo test`
 (integration tests `slhdsa_kat`, `scalar_hash_kat`); [MEASURE: run cargo test and publish the