⛲ SEI testnet faucet for EVM chains, Giga first.
It runs at https://faucet.giga-testnet-1.platform.sei.io for the Giga testnet
giga-testnet-1 (EVM chain ID 1330, RPC https://rpc.giga-testnet-1.platform.sei.io).
Its API, configuration and storage may still change without notice.
wellspring hands out testnet SEI on a Sei EVM chain. One process owns one signing key and its nonce, so every transfer is signed and sent in order, and a claim counts as paid only once its receipt succeeds.
It is testnet-only: it refuses to start for Sei mainnet (chain ID 1329).
make verify # lint, race tests and build, as CI runs them
make fmt # gofmt and goimportsmake e2e runs the binary against a real node: 31 claims, a SIGKILL mid-batch
and a restart, then checks every recipient was paid exactly once. CI runs it
against anvil. To run it against a local Giga (Autobahn) devnet whose genesis
funds the signer:
WELLSPRING_E2E_RPC_URL=http://localhost:8545 WELLSPRING_E2E_KEY_FILE="$PWD/key" make e2eEvery setting is a flag, and each flag can also be set from a WELLSPRING_*
environment variable: --rpc-url is WELLSPRING_RPC_URL. Run
wellspring --help for the full list and defaults.
| Flag | Default | Meaning |
|---|---|---|
--rpc-url |
required | EVM JSON-RPC endpoint; must serve --chain-id |
--chain-id |
required | EVM chain ID to sign for; 1329 (mainnet) is refused |
--chain-name / --public-rpc-url |
Sei testnet / none |
What the page shows and offers wallets |
--private-key-file |
required | File holding the hex secp256k1 signer key |
--amount-wei |
1 SEI | Amount per claim |
--max-recipient-balance-wei |
10 SEI | Refuse addresses already holding this much |
--min-faucet-balance-wei |
1000 SEI | Below this, /readyz fails |
--address-cooldown |
24h |
One claim per address per cooldown |
--ip-window / --ip-limit |
24h / 5 |
Claims per client IP per window |
--max-queued |
1000 | Refuse new claims once this many wait |
--trust-proxy-hops |
0 | Reverse proxies in front, for X-Forwarded-For |
--captcha-provider |
turnstile |
turnstile, hcaptcha, or none for local use |
--captcha-site-key / --captcha-secret-file |
required unless none |
Captcha site key and a file holding its secret |
--batch-interval / --batch-size |
500ms / 100 |
How often and how many claims to send |
--gas-tip-cap-wei / --gas-fee-cap-wei |
1 / 10 gwei | EIP-1559 fee caps |
--confirm-timeout |
30s |
Wait for a receipt before resyncing the nonce |
--db-path |
wellspring.db |
SQLite database; put it on a persistent volume |
--listen-addr / --metrics-addr |
:8080 / :9090 |
HTTP and Prometheus listeners |
--log-level |
info |
JSON logs to stdout at this level |
Secrets are only read from files, such as a mounted Kubernetes secret, so they never appear in the environment or the pod spec.
The image is distroless and runs as the nonroot user (UID/GID 65532). It
only writes to /data, so run it with a read-only root filesystem and a volume
there:
docker build -t wellspring .
docker run --read-only -v wellspring-data:/data -v ./secrets:/secrets:ro \
--add-host=host.docker.internal:host-gateway \
-e WELLSPRING_RPC_URL=http://host.docker.internal:8545 -e WELLSPRING_CHAIN_ID=1330 \
-e WELLSPRING_PRIVATE_KEY_FILE=/secrets/key -e WELLSPRING_CAPTCHA_PROVIDER=none \
-p 8080:8080 -p 9090:9090 wellspringOn SIGTERM it stops taking claims, stops the sender and closes the database.
Claims it had not settled are reconciled from the chain on the next start. /healthz is
liveness; /readyz also checks the RPC and the faucet balance.
The Helm chart in chart/ runs it the only safe way: a single-replica
StatefulSet, which never starts a second pod while the first is still shutting
down, since one signer owns the nonce, a PVC at /data, and a
fixed security context (UID/GID 65532, read-only root filesystem, no
capabilities, no service account token). Merges to main publish a snapshot
image and chart; a vX.Y.Z tag publishes a release, with the chart pinned to
its image by digest.
The chart never creates secrets. Create the signer key and captcha secret yourself, then install:
kubectl create secret generic wellspring-signer --from-file=private-key=./key
kubectl create secret generic wellspring-captcha --from-file=secret=./captcha-secret
helm install wellspring oci://ghcr.io/sei-protocol/charts/wellspring --version <version> \
--set config.rpc-url=http://fullnode:8545 --set-string config.chain-id=1330 \
--set config.captcha-site-key=<site key> --set captcha.existingSecret=wellspring-captchaconfig takes any flag by name as a string, such as config.amount-wei. Set
config.trust-proxy-hops to the number of proxies in front, or every claim
counts against the proxy's IP. See chart/values.yaml for the HTTPRoute,
ServiceMonitor, PrometheusRule and NetworkPolicy settings, and
RUNBOOK.md for what to do when an alert fires.