> ## Documentation Index
> Fetch the complete documentation index at: https://docs.pipedata.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Send data

> Write NDJSON into a pipe from stdin or accept it over HTTP.

Two ways to get records into a pipe: stream NDJSON from a shell with `pd push`, or expose an HTTP endpoint with `pd source http`.

## TL;DR

```bash theme={null}
# From stdin
cat data.jsonl | pd push orders

# Over HTTP
pd source http orders
# → prints an ingest URL; POST NDJSON or JSON to it with the workspace API key
```

## From stdin: `pd push`

`pd push <name>` reads NDJSON from stdin and writes it to the named pipe. The pipe is created on first use.

```bash theme={null}
$ echo '{"event":"signup","user":1}' | pd push events
$ kubectl logs -f my-svc             | pd push logs
$ cat backfill.jsonl                 | pd push orders
```

Each line of stdin must be a single JSON value. Status output goes to stderr, so redirects stay clean.

<Note>`pd push` is streaming — the writer doesn't wait for a reader. Push as fast as you like; readers catch up at their own pace.</Note>

## Over HTTP: `pd source http`

`pd source http <name>` creates (or looks up) an HTTP ingestion pipe and prints its ingest URL.

```bash theme={null}
$ pd source http webhooks
created source:http "webhooks"

  POST records to:
    https://ingress.eu-central-1.pipedata.io/123/data/456

  authenticate with your workspace API key:
    Authorization: Bearer <api-key>
```

Send one record per request, or batch many as NDJSON:

```bash theme={null}
# Single record
curl -X POST "$INGEST_URL" \
  -H "Authorization: Bearer $PD_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"event":"signup","user":1}'

# Batch (NDJSON, one record per line)
curl -X POST "$INGEST_URL" \
  -H "Authorization: Bearer $PD_API_KEY" \
  -H "Content-Type: application/x-ndjson" \
  --data-binary @batch.ndjson
```

A `2xx` response means the records were accepted and persisted.

## Common patterns

**Backfill a pipe from a file**

```bash theme={null}
$ cat history.jsonl | pd push orders
```

**Convert CSV on the fly**

```bash theme={null}
$ awk -F, 'NR>1 {printf "{\"email\":\"%s\"}\n",$2}' users.csv \
  | pd push emails
```

**Mirror an HTTP webhook into a pipe**

```bash theme={null}
$ pd source http shopify-webhooks
# point Shopify at the printed URL with your API key in the Authorization header
```

## See also

* [Receive data](/guides/receive) — read records back with `pd pull`.
* [Forward to a webhook](/guides/forward) — chain a pipe to an HTTP destination.
* [Limits](/reference/limits) — record size and request limits.
* [`pd push`](/reference/commands#pd-push), [`pd source http`](/reference/commands#pd-source-http) in the command reference.