Pulse · dated evidence example

Seven days you can check yourself.

This is a historical example from 4–10 October 2026, not a current health report. Days before 4 October are excluded; days after 10 October are outside this bundle. It lets a developer reproduce checks on saved observations before deciding whether that history is useful.

What is in the sample?

188 endpoint URLs (our four self-observed URLs excluded) observed across the window; 158 appear on all seven days.
171,204 unpaid samples from one observer. This capped registry sample is not a market census.
Two original signed captures (packaging instructions corrected on 16 October), the reference checker, external Bitcoin headers and the coverage review.

Download the reproducible packet · SHA-256 file manifest

Reproduce the result

Unzip the packet and inspect its README and Python checker. Create an isolated Python environment and install the pinned requirements. The three pinned packages are opentimestamps 0.4.5, python-bitcoinlib 0.12.2 and pycryptodomex 3.23.0. Installation needs network access; verification itself is offline. Unzip into an empty directory, then run these commands from it:

python3 verify_archive.py verify oct7/bundle.json --headers oct7/bitcoin-headers.json --as-of 2026-10-07 --from 2026-10-04 --to 2026-10-05
python3 verify_archive.py verify oct12/bundle.json --headers oct12/bitcoin-headers.json --as-of 2026-10-12 --from 2026-10-06 --to 2026-10-10

Expected: exit 0 with two usable days, then exit 0 with five usable days. The --as-of argument fixes the freshness assessment date. Running with today’s date reports this window as stale. These explicit dates reproduce historical assessments; they do not make old evidence fresh today. Original saved reports also cover wider windows with a pending latest day. For a new assessment, capture new evidence and use the actual UTC date.

What passed—and what did not get tested

The checker checks publisher authentication against its pinned key, daily record counts and roots, commitment-chain arithmetic, and timestamp proofs against Bitcoin headers obtained from a public explorer. It does not validate full Bitcoin consensus or independently establish the truth of the observer’s measurements.

Nothing here proves seller safety, uptime between probes, successful payment or delivery, credential validity, customers or revenue. x401 credential evidence is not included. Four own endpoint URLs are excluded from the coverage statistics.

Excluded history and recovery limits

56 legacy days remain mismatched. The 2 October transition and 3 October collector incident are also excluded. Original records are preserved. The first repaired day, 4 October, has 162 total records; its exact root and count were reproduced from an archive-only off-server backup. That is a one-day recovery check, not full disaster recovery.

Keep inspecting the original refusal example · Collection method and limitations

Try one verdict, if the sample is useful

A single call, not a subscription. The current advertised price is 0.03 USDC on Base. Read the live payment offer before approving a purchase; it is the authoritative quote.

Look before paying. Open the free five-row feed sample to see current fields and sample counts. These five rows are not a coverage lookup for your URL. Run a free current check for your exact public endpoint. A free check does not enrol it or create history.
Ask for the quote. The command below sends one unpaid GET to Pulse. HTTP 402 and a PAYMENT-REQUIRED header are expected; they are not a failed purchase. It cannot pay and does not contact the target endpoint through the paid handler.
Use your x402 client. If the quote and scope suit you, make the same GET with your existing compatible client. See the official buyer quickstart. Your wallet needs the offered asset on the offered network. Review the amount, asset, recipient and spending policy before authorising. Opening the URL in a browser does not complete payment.
Inspect what came back. Check HTTP status, your client’s settlement result, verdict.metrics.probes, firstProbeTs, lastProbeTs, flags and qualityLimitations. A settled payment or a high grade alone does not establish useful delivery. If it changed a real endpoint decision, tell us what changed. If it added nothing, that is useful feedback too.

curl --get --include --max-time 20 \
  'https://pulse.chainverdict.xyz/v1/verdict' \
  --data-urlencode 'url=https://YOUR_PUBLIC_API/your-endpoint'

Checked on 17 October 2026: exact scheme, amount 30000 atomic units (0.03 USDC), network eip155:8453, asset 0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913, recipient 0xAe2634E709c454f2720C65A0b2F9ba168e431842. Compare every field with the live offer; these dated values are not permission to pay a changed quote.

Replace the example URL with a public endpoint you are authorised to submit. Include the exact path and query it needs, but never credentials, wallet keys, tokens or private information. Pulse and the target service receive the request when you run a live check or purchase a verdict. Do not submit a literal route placeholder such as :input.

History is not guaranteed. An endpoint we have not observed may return only the new probe. A purchased verdict enrols its public URL for scheduled observations; this is not a monitoring SLA. The URL and its observations can enter public, retained daily exports. Only submit URLs you are comfortable making public. Recent probe history is capped at 500 samples per endpoint; daily streaks use a separate rollup. Sampled availability is not continuous uptime. A streak ends at the latest observed day, which can be stale or still in progress; it is not a count of independently verified completed days. The live verdict JSON is not the authenticated historical packet above, and it does not prove successful paid delivery, seller safety or x401 credential validity.

Read the response schema. The compact illustration below is invented, incomplete and not production evidence. Even a high grade with one probe is only a sample:

{
  "verdict": {
    "score": 99,
    "grade": "A",
    "metrics": {
      "probes": 1,
      "availability": 1,
      "firstProbeTs": 1792152000000,
      "lastProbeTs": 1792152000000
    }
  }
}

Actual responses also include flags, latency measurements, the latest probe, a daily streak and quality limitations. Check those fields rather than treating this illustration as a target result.

Streak calculation corrected on 17 October 2026: missing UTC days now break consecutive runs. Earlier returned streaks may have counted across gaps. Original observations and previously issued tokens are preserved; their signatures do not correct the earlier calculation. Read the method and correction.

A crypto settlement is not automatically reversed by an HTTP error. There is no automatic refund flow. If you see a settlement but an unusable response, contact us so we can investigate; do not repeatedly retry payments.

Integration trouble? Ask for trial help with the public URL, time in UTC and error text. Remove payment headers, credentials and private details. No wallet connection or account signup is needed to read this page or the free examples.

Would this change your workflow?

Keep your fresh probe and payment controls. Would dated history change which endpoint you investigate—or add nothing? We welcome a concrete counterexample.

Send feedback to ChainVerdict. Please share synthetic examples or public URLs only, not credentials or private captures.