Independent Proof Demonstration
SecLog is a trade ledger that emits cryptographic proofs any third party can check offline. No account, no API key, no access to our infrastructure. Pick a hardware profile below, run the verifier in your own browser, then flip one bit and watch the proof collapse.
What This Page Proves, Precisely
A SecLog proof bundle is a small JSON artefact — about 1 KB — plus one Ed25519 public key. Together they demonstrate four facts no amount of operator tampering can fake.
Hashing the record with BLAKE3 and walking the Merkle path reconstructs the block root byte-for-byte. Any change to the record breaks this.
The reconstructed root is signed with Ed25519 by the operator's published key. The signature cannot be forged without that key.
A chain of block roots leads from this block to a payload that was published to independent witnesses at a point in time and itself signed.
The four checks are interlocked. Flipping a single bit anywhere — record, proof, signature, chain — causes at least one check to fail. Partial tampering is not possible.
That the record is true. That's the job of the exchange feed this record was normalised from. SecLog proves durability, not correctness.
BLAKE3, Ed25519, and the public key. Nothing else. No SecLog code is touched during verification, in this browser or in the Python script.
Live Verifier
Each bundle below was produced by a real SecLog run on the indicated hardware. Pick a profile — the verifier fetches that bundle and checks it live in your browser using two audited libraries: @noble/hashes (BLAKE3) and @noble/ed25519.
Throughput Artefact
Every bundle ships a timing.json from a 10-million-record run on the hardware in its title. It pins the CPU, core count, kernel, partition layout, and durability mode — a TPS number without that metadata is meaningless, so we don't emit one. Swap profiles above to see the same engine on different silicon.
durability_mode below.Throughput Measurement (No Proof)
This measurement record captures baseline throughput on identical hardware before and after the Zig 0.13 → 0.16 upgrade. This is measurement data only, not a cryptographic proof bundle. The signed proof bundle is being prepared in parallel (brief F8e) once the witness publisher is wired.
Throughput Measurement (No Proof)
This measurement record shows that Zig 0.16 on Apple M2 Pro reproduces the published Zig 0.13 numbers within measurement noise. The Linux Ryzen "regression" above is Linux-specific, not a Zig codegen defect. This is measurement data only, not a cryptographic proof bundle.
seclog-proof-bundle-apple-m2-pro.zip (Zig 0.13 baseline).Reproduce Without This Page
Download the bundle and run the reference Python verifier on your own machine. It's 200 lines, pure standard crypto, and makes no network calls.
pip install blake3 pynaclunzip seclog-proof-bundle-<profile>.zip -d bundle/python3 verify_proof.py bundle/--tamper.timing.json.$ python3 verify_proof.py bundle/ SecLog Proof Verifier ================================================== Libraries: blake3, pynacl (libsodium) 1. Record hash + inclusion proof (12 levels): PASS 2. Signature (Ed25519): PASS 3. Anchor signature (Ed25519): PASS 4. Chain integrity (5 links): PASS ================================================== VERDICT: VALID Throughput: 10,000,000 records in 2122ms = 4,711,515 records/sec measured on: AWS Graviton4 (Neoverse V2) — EC2 c8g.8xlarge (ca-central-1) (aarch64, 32 cores, linux 6.12) config: partitions=8, trades_per_block=4096, durability=none
Content Boundaries
Everything on this page is shareable with anyone — regulators, prospects, the public. The following is discussed only under NDA, because it reveals either a non-obvious engineering lever, a non-public commercial choice, or information about a specific customer.
Next Step
We can produce bundles against your own data, under NDA, on short notice — a specific symbol, date range, counterparty-identifier redaction profile, or a benchmark on your reference hardware.
Request a custom bundle