Inspired by James Bach (1997): Do not count vanity numbers — make evidence-based assessments with 3-stage plain-language coverage milestones.
Not met 6, to be confirmed 2, met 0. Counts come from each repository's exit criteria.
EC-1Valid execution evidenceNot met 2, to be confirmed 0, met 0EC-2No unresolved S1Not met 2, to be confirmed 0, met 0EC-3Traceability matrixNot met 0, to be confirmed 2, met 0EC-4Approval still within its validity periodNot met 2, to be confirmed 0, met 0EC-1Valid execution evidencecex-market-data-quality-labEC-2No unresolved S1cex-market-data-quality-labEC-4Approval still within its validity periodcex-market-data-quality-labEC-1Valid execution evidencesanyoii.github.ioEC-2No unresolved S1sanyoii.github.ioEC-4Approval still within its validity periodsanyoii.github.ioBy priority, the homepage shows S1 only.
S1cex-market-data-quality-labApproval expiredArchived display. Do not re-approve in this revision; retest after the revamp launches.S1sanyoii.github.ioApproval expiredArchived display. Do not re-approve in this revision; retest after the revamp launches.Known issues: no confirmed S1 in the pinned runs. Execution: historical Pass, evidence expired. Exit: self-review only, now expired, not an independent sign-off. Untested scope remains. Order is Blocked, At Risk, Unknown, Ready, so the current line is Unknown.
Instead of vanity percentages (e.g. 98%), we use 3 plain-language milestones: ① Basic Sanity → ② Edge & Negative Cases → ③ High-Load Stress. High-risk private orders and load tests are honestly disclosed as Excluded.
No unresolved S1 in the locked snapshots. The public Binance feed stays a warning because it depends on an outside network. Preview the S1 Fail hold to see why a confirmed Fail cannot be Ready.
Showing 6 Product AreasEvidence-based human assessment over vanity test counts — 3-stage plain-language milestones show exactly what is covered and what is excluded.
| 1. Product Area (Risk Domain) | 2. Effort | 3. Coverage Depth (3-Stage Milestones) | 4. Quality | 5. QA Assessment & Evidence |
|---|---|---|---|---|
CEX · Order Book Sequence & State RISK-OB-SEQ · Stale book must not overwrite amounts | High · Done | ✓ Basic Sanity✓ Edge & Negative◌ Stress Excluded | Good | Sequential updates, out-of-order rejection, and reconnect snapshot recovery all pass (TC-017 failure triggers immediate Hold).12 cases ↗ |
CEX · REST & WebSocket Contracts RISK-CONTRACT · Schema, heartbeat, and malformed payloads | High · Done | ✓ Basic Sanity✓ Edge & Negative◌ Stress Excluded | Good | Offline deterministic checks verify normal subscriptions, HTTP error codes, and malformed payload guards.23 cases ↗ |
CEX · Binance Public Live Feed RISK-LIVE · Public exchange endpoint, outside this repo | Med · Done | ✓ Basic Sanity▲ Public Feed Only◌ Chaos Excluded | Warning (Bounded) | Live BTCUSDT REST/WS handshake passed; subject to external network & exchange rate limits (no offline guarantee).5 live ↗ |
Portfolio · Bilingual & Chromium RWD RISK-I18N · EN/ZH parity and desktop/mobile viewports | Med · Done | ✓ Basic Sanity✓ EN/ZH & RWD◌ Safari Excluded | Good | Static HTML + Playwright runtime checks pass; honestly retains Attempt 1 environment-blocked record.22 checks ↗ |
Cross-Repo · Claim & Evidence Integrity RISK-DOC · Public claims stay inside the tested scope | Low · Done | ✓ Basic Sanity✓ Digest Audit◌ — | Good | All public claims are pinned to Git SHAs and sanitized JUnit SHA-256 digests without inflated percentages.5 checks ↗ |
Explicit Exclusions · Private Keys / Load / Safari Honest Boundary (Explicit Out-of-Scope · Never claimed as tested) | None · OOS | 🚫 Explicitly Excluded (Untested · Not a Zero-Risk Claim) | Untested (Disclosed) | Intentionally excludes live fund orders and production load testing; green checks elsewhere never imply full-system zero risk.Scope G1 ↗ |
00–02What must not break vs. explicit exclusions
4 TrackedRISKOrder Book Sequence & State IntegrityCoveredRISKREST & WebSocket Schema ContractCoveredRISKPublic Binance Live Feed HandshakeCoveredRISKDocumentation & Claim Boundary SyncCovered03–04Deterministic suites, live checks & attempt logs
40 / 40 PASS12/12 PASS11/11 PASS12/12 PASS5/5 PASS5 / 5 PASS3/3 PASS2/2 PASSVERIFIED05–06Current blocker state + proof the gate stops bad builds
0 BLOCKERSNO-GO PROOF[DEF-MDQ-004 · S1 demo] TC-MDQ-017 failed: stale sequence 103 overwrote sequence 104 in order-book state.
[Replay Rule] A hypothetical Pass would be a new Run plus RISK-OB-SEQ regression. It is not a Go, and this demo does not record that Run.
07–08Evidence-backed recommendation & TTL boundary
4/4 Met| Repository / Snapshot | Pinned Verdict | 1. Unit / Static | 2. Contract / Runtime | 3. Live / RWD | 4. Triage / Chronology | 5. Risk Coverage | 6. Exclusions | Evidence |
|---|---|---|---|---|---|---|---|---|
cex-market-data-quality-lab @fc1e060Scoped automated market-data acceptance | Unknown | ● 12/12 PASS | ● 28/28 PASS | ● 5/5 LIVE | ● 4 RUNS OK | ● CORE & EDGE | ◌ AUTH/LOAD | #RUN-20260907 ↗ |
sanyoii.github.io @643b017Primary QA portfolio static & Chromium acceptance | Unknown | ● 9/9 STATIC | ● 13/13 RUNTIME | ● EN/ZH + RWD | ▲ 1 BLOCKED→PASS | ● CORE & EDGE | ◌ NON-CHROMIUM | #RUN-20260906 ↗ |
cex-market-data-quality-lab (Hold Sim) @DEMO-FAILCounterfactual S1 Fail; not a real sign-off | ■ HOLD / NO-GO | ■ TC-017 FAIL | ● 2/3 PASS | ▲ STALE SEQ | ■ DEF-MDQ-004 | ▲ HOLD AT G6 | ◌ SYNTHETIC | Fail Demo ↗ |
67 CHECKS · 34 CASESSanitized JUnit SHA-256 telemetry matrix for Sept 6–7 releases plus interactive 34-case specification workbench (OB, REST, WS, DOC, LIVE, MTC).
Open Run Records Workbench →DEF-MDQ-004 · S1 demo4-stage counterfactual incident pipeline for TC-MDQ-017, interactive 3-step Hold boundary simulator, Expected/Actual diff, and Exit Code 1 pytest log.
Launch Failure & Gate Replay →LOW-TECH v3 · 6 RULESArchitecture evolution from early cards/matrices to the Low-Tech Whiteboard v3 and the 6 deterministic fail-closed remediation rules.
Inspect Architecture & Policy Matrix →