# Omnistudio Asynchronous Data Operations

> Use Integration Procedures queues, DataRaptor Chain, and Remote Actions with async patterns for long-running OmniStudio flows. NOT for simple DataRaptor reads — use omnistudio/dataraptor-patterns.

- Skill: `pranavnagrecha/omnistudio-asynchronous-data-operations` (Agent Skill, multi-file: 7 files)
- Install (CLI): `npx skillmds add pranavnagrecha/omnistudio-asynchronous-data-operations`
- Raw SKILL.md: https://api.skillmd.com/api/skills/pranavnagrecha/omnistudio-asynchronous-data-operations/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Integrations & APIs
- Author: PranavNagrecha (https://skillmd.com/u/pranavnagrecha)
- Updated: 2026-09-08
- Page: https://skillmd.com/skills/pranavnagrecha/omnistudio-asynchronous-data-operations

---


# OmniStudio Asynchronous Data Operations

OmniStudio Integration Procedures (IPs) can chain DataRaptors, HTTP calls, Apex, and Business Rules into a single orchestration. For operations that exceed the 5-second browser wait or need parallel calls, queue the IP (invoke from Platform Event or Queueable) instead of running synchronously.

## Adoption Signals

Multi-system orchestrations >5s or bulk data transformations driven from an OmniScript.

- Any IP that exceeds the synchronous CPU limit when run against production-volume data.
- DataRaptor chains (sometimes called a dataraptor chain) where each step depends on the prior output and total runtime exceeds user-tolerable latency.

## Recommended Workflow

1. Model the orchestration as an IP; label each step by purpose.
2. For long steps, split: a synchronous IP kicks off a Queueable or publishes a Platform Event; the worker IP runs async.
3. Consumer polls status via a DataRaptor read; show progress screen in the OmniScript.
4. Error path: each IP step has Abort On Error; always catch at top with a retry/fail branch.
5. Observability: log IP execution via custom logger or Salesforce Log Entry framework.

## Key Considerations

- The "120s overall, 10s per callout" figures are **Apex callout limits**, not OmniStudio settings — they bind an IP because its HTTP action runs in an Apex transaction. Default is 10s per callout (settable from 1 ms to 120,000 ms), with 120s cumulative across all callouts in the transaction. There is no OmniStudio screen for either.
- Cache responses where safe (cache-enabled IP) to reduce backend load.
- Parallel Remote Actions reduce wall-clock; IP supports Send/Response parallel.
- Don't put heavy transforms in DataRaptor — use Apex Action for perf-sensitive paths.

## Worked Examples (see `references/examples.md`)

- *Async order placement* — OmniScript checkout
- *Parallel enrichment* — Credit + address verification

## Common Gotchas (see `references/gotchas.md`)

- **Cumulative callout timeout exceeded** — the Apex 120s per-transaction ceiling aborts the whole IP.
- **Cache stale** — Users see old data.
- **No error branch** — Silent failure to user.

## Top LLM Anti-Patterns (full list in `references/llm-anti-patterns.md`)

- Everything sync
- No caching on hot IPs
- Heavy Apex in DataRaptor transformations

## Official Sources Used

- OmniStudio Developer Guide — https://developer.salesforce.com/docs/atlas.en-us.omnistudio_developer.meta/omnistudio_developer/
- Callout Limits and Limitations (10s default / 120,000 ms maximum per callout, 120s cumulative per transaction) — https://developer.salesforce.com/docs/atlas.en-us.apexcode.meta/apexcode/apex_callouts_timeouts.htm
- OmniStudio for Salesforce — https://help.salesforce.com/s/articleView?id=sf.os_omnistudio_for_salesforce_overview.htm
- OmniScript to LWC OSS — https://developer.salesforce.com/docs/atlas.en-us.omnistudio_developer.meta/omnistudio_developer/os_migrate_from_vf_to_lwc.htm