PranavNagrecha/AwesomeSalesforceSkills / skills/admin/flow-orchestration-admin
flow-orchestration-admin
VerifiedFlow Orchestration admin: stage configuration, step assignment, background steps, interactive steps, evaluation flows, work items, pause/resume. NOT for approval routing a standard approval process would handle — use admin/approval-processes.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 6.6 KB |
| references/examples.md | markdown | 6.6 KB |
| references/gotchas.md | markdown | 8.9 KB |
| references/llm-anti-patterns.md | markdown | 4.5 KB |
| references/well-architected.md | markdown | 5.9 KB |
| scripts/check_flow_orchestration_admin.pyexecutable | script | 4.0 KB |
| templates/flow-orchestration-admin-template.md | markdown | 635 B |
references/llm-anti-patterns.md
View on GitHub# LLM Anti-Patterns — Flow Orchestration Admin
Common mistakes AI coding assistants make when configuring Flow Orchestration.
## Anti-Pattern 1: Using Approval Process for multi-actor parallel work
**What the LLM generates:** A standard Approval Process with multiple steps for what is clearly a multi-team parallel review.
**Why it happens:** Approval Process is well-documented and familiar; Flow Orchestration is newer.
**Correct pattern:**
```
Parallel steps with independent actors = Flow Orchestration Stage with
multiple parallel Interactive Steps. Approval Process steps run in series
and cannot branch.
```
**Detection hint:** An Approval Process with three "unanimous required" steps that the business describes as "happening at the same time."
---
## Anti-Pattern 2: Heavy logic inside Evaluation Flows
**What the LLM generates:** A 15-element autolaunched flow with SOQL, loops, and decision branches as the Stage's evaluation flow.
**Why it happens:** Evaluation Flows look like regular flows; the cost of running them repeatedly is not obvious.
**Correct pattern:**
```
Evaluation Flow = one or two decisions returning a boolean. Pre-compute
context upstream (in a Background Step), then the Evaluation Flow just
reads a field and returns true/false.
```
**Detection hint:** An Evaluation Flow over ~5 elements or containing loops.
---
## Anti-Pattern 3: Interactive Step with no assignment
**What the LLM generates:** An Interactive Step that produces a Work Item but does not specify an assignee — "it will default to the current user."
**Why it happens:** The model does not realize Work Items need explicit assignment to surface on a user's list.
**Correct pattern:**
```
Every Interactive Step assigns to a specific user, queue, or evaluated
field. Unassigned work items get lost in the global list; users never
see them.
```
**Detection hint:** A Flow Orchestration metadata XML with an Interactive Step missing assignee fields.
---
## Anti-Pattern 4: Rolling a long-wait via scheduled Apex
**What the LLM generates:** Suggests a scheduled Apex to poll an external system and advance the orchestration.
**Why it happens:** The model falls back on polling patterns from non-orchestration training data.
**Correct pattern:**
```
Use a Platform Event from the external system. The Background Step waits
on the event. The external system publishes on completion; orchestration
resumes. No polling, lower cost, tighter coupling to the real event.
```
**Detection hint:** Scheduled Apex that advances orchestrations by updating records or invoking orchestration APIs.
---
## Anti-Pattern 5: Deleting orchestration metadata while instances are running
**What the LLM generates:** "Retire the old orchestration — delete the metadata, deploy the new one."
**Why it happens:** The model treats orchestration like any flow.
**Correct pattern:**
```
Drain in-flight instances first (either complete them, cancel them, or
migrate). Deploy the new orchestration alongside the old. Only delete
old metadata after the monitoring view confirms zero active instances.
```
**Detection hint:** A retirement plan that deletes Orchestration metadata on the same change window as go-live.
---
## Anti-Pattern 6: Sending ops to Setup → Orchestration Runs
**What the LLM generates:** "Open Setup, search Orchestration Runs."
**Why it happens:** Paused Flow Interviews live in Setup, so the analogue is guessed.
**Correct pattern:**
```
/lightning/o/FlowOrchestrationInstance/list
```
That is the object list view. `/lightning/setup/OrchestrationRuns/home` is not a Setup node.
**Detection hint:** Any URL containing `setup/OrchestrationRuns`.
---
## Anti-Pattern 7: A "just click Resume" runbook that never names the run's status
**What the LLM generates:** "If the run is stuck, click Resume."
**Why it happens:** One control name covers more than one run state, so a single instruction looks complete.
**Correct pattern:** Name the state first — a run stalled on an unclaimed work item, a run that errored in a step, and a run paused waiting on something outside Salesforce are three different situations with three different operator actions, and only the last is resumed by publishing a `FlowOrchestrationEvent`. Read the exact resume semantics for each state off the Orchestration Runs detail view in the target org and write them down; do not state them from memory in a runbook someone will follow at 2am.
**Detection hint:** A single "resume" procedure that does not name the run's current status.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack." | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |