API reference
Get Started

Operations / Running the demo

Running the demo

The whole flow on a local Surfpool network, in one command.

Run it

brew install txtx/taps/surfpool      # Surfpool 1.0
docker compose up -d                  # Postgres
pnpm --filter @corpact/demo demo      # about 4½ minutes; exits 1 if any check fails

The run writes walkthrough.md and report.json to apps/demo/out/<run>/. Set DEMO_VALIDATOR=solana-test-validator to run on the Agave test validator instead.

What you see

Part 1: Walkthrough (synthetic)

  1. Position. A holder of 100 units: raw base units, displayed quantity and protected floor.
  2. No-transfer dividend. A scheduled multiplier change activates. The raw balance and the account's transaction count stay the same, and the displayed quantity rises.
  3. Dividend entry. 0.35 extra shares, worth $35.00 from issuer net cash, retained in stock.
  4. Split. 2-for-1: the quantity and the floor double, and income stays $35.00.
  5. Correction. The issuer revises withholding. The journal keeps the original, appends a reversal and a $33.00 replacement, and pauses conversion.

Part 2: Recorded cases

Six cases from recorded issuer data, side by side with the naive reading: the HONx spin-off, STRCx implausible cash, the KRAQx rights sale labelled UnitSplit, the SCCOx stock dividend with six versions, the LINx withholding refund, and the AZNx ADR conversion labelled "ReverseSplit".

Part 3: Trap regression suite

Every Phase 0 trap the ledger can reproduce without a price source runs end to end:

  • missing and implausible cash;
  • a spin-off labelled "Dividend";
  • a superseded schedule, a late publication and a late issuer record;
  • partial history;
  • deterministic replay;
  • an independent provider that disagrees, then agrees again.

Synthetic data is labelled everywhere

  • Network. Surfpool runs offline, with keys generated per run.
  • Database. corpact_demo is permanently labelled synthetic, and every response, header and CSV row says so. See Synthetic vs mainnet data.