Dash Platform QA Dashboard

A read-only status dashboard for the on-chain QA framework on Dash Platform. It reads the dash-qa data contract (document types testCase and testRun), computes the latest result per test client-side, and renders a tier × category status matrix, per-test run history, summary counts, and filters.

• Read-only (v1): no wallet, no signing, no writes. Queries are proof-verified via the Evo SDK (trusted mode pre-fetches quorum keys and verifies Platform proofs).
• Stack: Vite + React + TypeScript.
• Data source: the dash-qa contract produced by the sibling "on-chain QA data contract + seed from TEST_PLAN" task. The contract ID and network are fully configurable (see Configuration).

Submitting runs from the browser (which needs wallet signing) is intentionally out of scope for v1. The seam is stubbed under src/features/submit-run/ so it can be added later without touching the read path.

Quick start

npm install
npm run dev        # http://localhost:5173

Open the app and either:

• click Load demo data (or visit ?demo=1) to explore with a bundled sample dataset — no network needed; or
• open ⚙ Data source, paste the dash-qa contract id, pick the network, and click Apply & load.

Configuration

The dashboard needs a contract id and a network. There is no single hard-coded value — resolution runs through these layers, highest precedence first:

Precedence	Source	Use case
1	URL query — ?contract=<id>&network=testnet, or ?demo=1	Shareable deep links (e.g. from TEST_PLAN / a QA issue).
2	In-app settings (⚙ Data source) → localStorage	A person trying it locally.
3	public/config.json (fetched at runtime)	A fixed deploy you can re-point without rebuilding.
4	Build-time env — VITE_CONTRACT_ID, VITE_NETWORK	Baking the contract into the bundle / CI.
5	Defaults (network: testnet, contract empty)	—

Where to get the contract id

The sibling contract task writes contract-id.<network>.json (e.g. contract-id.testnet.json) when it registers the contract:

{ "network": "testnet", "contractId": "<base58 contract id>", "...": "..." }

Copy that contractId into whichever layer fits:

• Local: paste it into ⚙ Data source, or copy .env.example → .env and set VITE_CONTRACT_ID.
• Fixed deploy without rebuild: copy public/config.example.json → public/config.json and set contractId (this file is gitignored by default; commit it if you want it version-controlled).
• CI / GitHub Pages: set the CONTRACT_ID (and optional NETWORK) repository variables — the deploy workflow passes them as VITE_*.

Testnet resets periodically, so the contract id changes when it is re-registered. Re-point the dashboard via any layer above; no code change needed.

Deep links

• ?demo=1 — load the bundled demo dataset.
• ?contract=<base58>&network=testnet — point at a specific contract.
• ?test=CORE-05 — open that test's run-history drawer on load.

Expected contract schema

The dashboard reads these dash-qa document types (field names mirror the sibling task's schema/qa-contract.documents.json). The normaliser (src/sdk/normalize.ts) is tolerant of synonyms and of the TEST_PLAN emoji vocabulary, so minor drift degrades gracefully.

The contract is normalised (v3+): tier, category, and app are separate lookup document types ({ code, name }), and testCase / testRun reference them by integer code. The dashboard fetches those lookups up front and resolves the codes back to names (so the matrix/filters show Essential / Core, not 0 / 0); unresolved or string-typed values pass through, so older contracts still work.

Lookups — tier / category / app, each { code (int), name } (app also carries platform, description).

testCase — one row of the TEST_PLAN §4 catalog: testId (e.g. CORE-05), app/tier/category (integer FK codes → resolved names), title, layer, implStatus (the glyph ✅/🧪/⚠️/🔌/🚫), description, entryPoint, prerequisites, planCommit.

testRun — an append-only execution record: testId, app (FK code → name), result (enum pass/fail/blocked/skipped), network (integer id — 0=mainnet, 1=testnet, 2=devnet, 3=regtest — mapped to a name), buildRef, device, evidence (txid / URL / path — classified for display), notes, blockerReason. $createdAt is the run time. The unique (testId, app) index allows per-app results; with a single seeded app today the dashboard keys "latest result" by testId.

How "latest result" is computed

Platform has no GROUP BY, so the reduction is client-side: fetch all testRun docs (paginated), sort by $createdAt desc, then take the first run per testId. The contract indexes $createdAt only inside $ownerId-prefixed composite indices (e.g. [$ownerId,testId,$createdAt]), so a global "order all runs by time" query isn't possible — paging through every run and sorting in the client is both index-safe and complete for an append-only audit log of this size. Runs are loaded up to a cap (DEFAULT_MAX_DOCS, 5000); if exceeded, only the oldest beyond the cap are dropped.

Scripts

npm run dev        # dev server (serves from / locally)
npm run build      # typecheck + production build to dist/
npm run preview    # serve the production build locally
npm test           # unit tests (vitest) — normalise + aggregation logic
npm run typecheck  # tsc, no emit

Deployment (GitHub Pages)

.github/workflows/deploy.yml runs on push to main: it typechecks, tests, builds, and publishes dist/ to GitHub Pages.

The production build uses base path /qa-dashboard-site/ (the repo name) to match https://<org>.github.io/qa-dashboard-site/. For a custom domain or Vercel, set VITE_BASE=/ at build time.

One-time setup: in the repo settings, enable Pages → Build and deployment → Source: GitHub Actions. Optionally set the CONTRACT_ID repository variable to bake the live contract into the deployed bundle.

Project structure

src/
  config.ts              # contract id + network resolution (URL/LS/json/env)
  sdk/
    client.ts            # Evo SDK connection (trusted, memoised per network)
    documents.ts         # fetch testCase / testRun (+ demo short-circuit)
    normalize.ts         # raw Platform docs → typed model (tolerant)
    types.ts             # domain model + enums + ordering
  data/
    compute.ts           # latest-per-test, matrix, summary, filters (pure)
    useQaData.ts         # React hook: load + reload
    demo.ts              # bundled demo dataset (?demo=1)
  components/            # SummaryCounts, StatusMatrix, Filters, TestList, RunHistory, …
  features/submit-run/   # future write feature — stub only, never imported by reads

License

MIT