Testnet programme — plans, awaiting approval

Master Prompt 19 §4-5. Nothing here has been launched, provisioned or announced. Each phase starts only after an explicit “APPROVED: ”. An infracost estimate is required before each approval request, and no estimate exists yet (reports/19-operations.md).

Staging network (plan)

Item	Plan
Nodes	50 full nodes: 40 validators (once DAG-BFT is wired) + 10 RPC
Regions / providers	at least 3 regions on 2 providers (the terraform in infra/terraform targets one provider today, so a second is to be written)
Size per node	docs/NODE_TYPES.md validator row: 4 vCPU, 8 GiB, NVMe, 1 Gbit/s
Purpose	every release candidate soaks here before any public release; every SLO in docs/SLO.md is measured here
Approval	“APPROVED: staging-network”

Phase 1 — public testnet

Open node running, faucet (apps/faucet), explorer (apps/explorer), status page. Exit criteria: 30 days without a chain halt; cargo xtask slo-check green; the conformance suite green on every release. Approval: “APPROVED: public-testnet”.

Phase 2 — attacknet

Invited researchers get validator slots and a bounty for halting, forking or censoring. Rules of engagement:

• in scope: the chain, its p2p and RPC surfaces, the signer channel;
• out of scope: third-party infrastructure, social engineering, anything touching real funds;
• disclosure through SECURITY.md.

Publishing the bounty needs “APPROVED: bug-bounty”; running the attacknet needs “APPROVED: attacknet”.

Phase 3 — incentivized testnet

External validators across many countries and providers, with uptime and performance tracked. The rewards policy is drafted for legal review (docs/legal/QUESTIONS_FOR_COUNSEL.md: token classification of testnet rewards). Approval: “APPROVED: incentivized-testnet”, after counsel.

Load events

Scheduled public stress tests with maya2c-loadgen and maya2c-rpc-load at increasing levels. Results are published with the methodology in docs/BENCHMARK_METHODOLOGY.md so anyone can reproduce them.