Task functions
A workflow is an ordered list of tasks, and every task invokes one built-in function with an input object. Functions read and write the data context, the JSON document that flows through the pipeline. This hub lists every function the release documented here ships; each name links to its own page.
Some functions are contributed by the dataflow-rs engine. The rest are Orion handlers that talk to connectors, compose channels, or compute locally. GET /api/v1/admin/functions serves the authoritative list on a running instance; see Runtime function discovery.
| Function | Category | Connector | Purpose |
|---|---|---|---|
parse_json | Data | — | Parse the raw payload into the data context |
parse_xml | Data | — | Parse an XML payload into the data context |
map | Data | — | Transform/reshape data with JSONLogic |
filter | Data | — | Gate the pipeline on a JSONLogic condition |
validation | Data | — | Collect validation errors from JSONLogic rules |
log | Data | — | Emit a structured log line |
publish_json | Data | — | Serialize a context field to a JSON string |
publish_xml | Data | — | Serialize a context field to an XML string |
http_call | Connector | HTTP | Call an external API with retry + circuit breaker |
data_query | Connector | SQL / MongoDB / ES | Portable, backend-neutral query |
data_write | Connector | SQL / MongoDB / ES | Portable, backend-neutral insert/update/delete/upsert |
db_read | Connector | SQL | Run a raw SELECT, return rows as JSON |
db_write | Connector | SQL | Run raw INSERT/UPDATE/DELETE, return affected count |
cache_read | Connector | Cache | Read a value from Redis or the in-memory cache |
cache_write | Connector | Cache | Write a value to cache with optional TTL |
mongo_read | Connector | MongoDB | Run a raw find(), return documents as JSON |
mongo_write | Connector | MongoDB | Insert/update/replace/delete documents, nested shapes included |
mongo_aggregate | Connector | MongoDB | Run a stage-allowlisted aggregation pipeline |
publish_kafka | Connector | Kafka | Publish a message to a Kafka topic |
send_email | Connector | SMTP | Send transactional email through an SMTP connector |
storage_presign | Connector | Storage | Compute a time-limited presigned object URL — no data path |
storage_head | Connector | Storage | Object metadata (exists/size/etag) |
channel_call | Composition | — | Invoke another channel’s workflow in-process |
model_infer | Compute | — | Run an admitted ONNX model: the manifest’s adapters in, tensors through, the result out |
crypto | Utility | — | Digests, HMAC compute/verify, password hashing |
jwt_sign | Utility | — | Mint a signed JWT (login, refresh, client assertions) |
jwt_verify | Utility | — | Verify a JWT against static keys or a JWKS |
Note
The Category column groups the table for reading. It is not the wire value:
GET /api/v1/admin/functionsserves acategoryofconnector,control,data,compute, orutilityfor every function, so tooling should branch on those rather than on the labels here.
When a function fails, the data-plane response follows Errors and response envelopes. Use its stable error code and the trace’s task ID to identify the failing step; do not parse the message text. Each function’s page describes its own validation and runtime failures.
Related
- Workflows: the pipeline model every function runs in.
- Author a workflow: choosing and combining functions in practice.
- Workflow definition: the task object a function sits in, and the data context.
- Retry safety: what a retry of each function costs.
- Runtime function discovery: the live catalogue, plugin functions included.
Last verified 14 September 2026