Examples

Recipes

Each recipe is a complete pattern you can adapt. Rules and URLs are illustrative; set limits from your provider's documented quota.

fetch, undici, axios, and got, with provider feedback.

Process a large list within a quota, with progress and failures.

Share one quota across BullMQ worker processes.

Call a provider from a web server and fail fast when busy.

One valve per customer credential, created on demand.

Per-second and daily limits, with weighted endpoints.

The repository includes small, self-checking programs under examples/. Each asserts its own result and needs no provider credentials.

Example	Shows	Needs Redis
`memory`	Spacing calls in one process with the default memory store.	No
`priorities`	Interactive work starting before background work.	No
`observe`	Feeding an axios-shaped response into shared feedback.	No
`shutdown`	Rejecting queued work and finishing running work on `close`.	No
`shared`	Six valve instances sharing one Redis budget.	Yes
`feedback`	A 429 in one valve pausing another valve on the same budget.	Yes

Build the workspace once, then run any example by name:

pnpm install
pnpm build
pnpm --filter @valve/example-memory start

For the Redis examples, start a server and set REDIS_URL:

docker compose -f compose.test.yml up -d
REDIS_URL=redis://127.0.0.1:6379 pnpm --filter @valve/example-feedback start

pnpm examples runs all six in sequence.