service-omni-queue-routing-config-deploy

SkillCommunication

Lets your agent use an omni claude skill to create or update Salesforce queue routing configurations.

Available today. Use it from your connected AI after setup.

Add ahel to your AI once: Claude, ChatGPT, Cursor, Claude Code or Codex. Then ask it to use this.

Then ask your AI: use the service-omni-queue-routing-config-deploy skill

About this skill

Use to create or update a QueueRoutingConfig record on a Salesforce org via the Data API (idempotent SOQL detect + REST upsert). Naming is target-derived: Case→Case_Routing_Config, VoiceCall→Voice_Routing_Config, Incident→Incident_Routing_Config, MessagingSession→Messaging_Routing_Config (via QRC_RO

What this skill tells your AI

The instructions your AI receives, as published by forcedotcom/sf-skills in skills/service-omni-queue-routing-config-deploy/SKILL.md and read by ahel’s review.

Create or update a QueueRoutingConfig via the Data API. This record tells Omni-Channel how to distribute work to agents on a queue: the routing model (MostAvailable for round-robin-style, LeastActive for load-balancing), how much capacity each work item consumes (CapacityWeight or CapacityPercentage), the push timeout, and the routing priority. It is a prerequisite for an Omni-routed Queue — service-omni-queue-deploy binds the queue to the config Id this skill returns. Detection is by DeveloperName, so the skill POSTs a new record or PATCHes an existing one to align, and is safe to re-run.

Inputs

bash scripts/upsert-and-report.sh <org-alias> [routing-model] [developer-name] [master-label] [capacity-weight] [routing-priority] [overflow-assignee] [push-timeout]
# Case_Routing_Config, preserve model / create MostAvailable:
bash scripts/upsert-and-report.sh myorg
# align the Case QRC to LeastActive:
bash scripts/upsert-and-report.sh myorg LeastActive
# Voice_Routing_Config (coordinator path):
QRC_ROUTING_TARGET=VoiceCall bash scripts/upsert-and-report.sh myorg
# set a fallback/overflow assignee (route overflow work to a supervisor):
QRC_OVERFLOW_ASSIGNEE=supervisor@example.com bash scripts/upsert-and-report.sh myorg
# Voice uses 100% capacity and a 30-second push timeout:
QRC_ROUTING_TARGET=VoiceCall QRC_CAPACITY_PERCENTAGE=100 QRC_PUSH_TIMEOUT=30 bash scripts/upsert-and-report.sh myorg MostAvailable
  • org-alias (required).
  • routing-model (optional, or ROUTING_MODEL_INPUT) — MostAvailable | LeastActive. Omitted: an existing record's model is preserved; a new record is created MostAvailable. Provided: written on create, aligned on an existing record.
  • developer-name / master-label / capacity-weight (1..100) / routing-priority (≥1) — positional after routing-model, or via QRC_DEVELOPER_NAME / QRC_MASTER_LABEL / QRC_CAPACITY_WEIGHT / QRC_ROUTING_PRIORITY. Label, weight 5, and priority 1 are defaults only when creating a new record; omitted values preserve an existing record.
  • QRC_CAPACITY_PERCENTAGE (optional, 0..100) — use percentage capacity instead of weight. It is mutually exclusive with an explicitly supplied capacity weight; switching modes clears the old sibling field.
  • overflow-assignee (optional 7th positional, or QRC_OVERFLOW_ASSIGNEE) — the fallback User (Username or 005 Id) that receives work when every queue member is unavailable (OverflowAssigneeId). Resolved to an active User Id before the write; omitted leaves an existing record's overflow untouched and a new record with none.
  • push-timeout (optional 8th positional, or QRC_PUSH_TIMEOUT, 0..3600 seconds) — how long Omni waits for an agent to accept pushed work. Omitted preserves an existing value.
  • QRC_ROUTING_TARGET (env) — Case | VoiceCall | Incident | MessagingSession; selects the canonical naming defaults. service-omni-channel-setup-coordinate sets this per target.

Preconditions and safety

  • Target org authenticated via sf CLI (My Domain URL), Service Cloud license, sf CLI ≥ 2.139.6.
  • Omni-Channel base settings enabled (service-omni-base-settings-configure) — otherwise QueueRoutingConfig is not writable (INVALID_TYPE).
  • The three-way safe_to_write production guard applies.

Run

upsert-and-report.sh performs detect → upsert → verify:

  1. Query QueueRoutingConfig WHERE DeveloperName = '<target-derived-or-explicit>', capturing the existing Id and field values.
  2. Compare against the requested/canonical values.
  3. POST a new record when none exists; no-op (reused) when all fields already match; PATCH when a field differs.
  4. Re-query to confirm the final state; block if the re-query shows a mismatch.

Behavior

Target-derived naming. With no explicit name, the DeveloperName follows the routing target (Case_Routing_Config, Voice_Routing_Config, etc.), so re-runs and the coordinator converge on the same record rather than creating parallel duplicates.

Preserve over churn. On an existing record, omitted routing model, label, routing priority, capacity mode/value, overflow assignee, push timeout, and attribute-based flag are preserved. Defaults (MostAvailable, target-derived label, weight 5, priority 1, non-attribute-based) apply only to a brand-new record. Capacity uses exactly one mode; explicitly changing from weight to percentage, or back, clears the old sibling field so the record cannot contain conflicting capacity settings.

Fallback / overflow routing. In the QueueBased model, "fallback" is the queue's overflow assignee — the User who receives work when no queue member is available (rather than the item sitting queued indefinitely). Agent availability is an async Omni decision, not observable in a routing flow, so this belongs on the QueueRoutingConfig, not the flow. When overflow-assignee is provided the skill resolves it to an active User (blocking on an unknown or inactive user, since overflow to an inactive user would silently strand work), writes OverflowAssigneeId, and verifies it round-trips.

Read-back reporting. All reported field values are re-queried from the org after the write (values_source: "requeried"), not echoed from inputs — a 201/204 only means the write was accepted. The skill never deletes a QueueRoutingConfig (queues, agents, and work skills may reference the Id) — it always PATCHes to align.

Output contract

A single JSON object with status ∈ created | reused | updated | blocked, plus routing_target, developer_name, id, master_label, routing_model, routing_priority, capacity_mode, capacity_weight, capacity_percentage, is_attribute_based, push_timeout, overflow_assignee_id, overflow_assignee_requested, values_source, a before snapshot, manual_actions, and blocking_issue.

  • created — no record existed; POST succeeded.
  • reused — a record existed with all fields matching; no-op.
  • updated — a record existed but a field differed; PATCH succeeded.
  • blocked — precondition failed (Omni not enabled, production org, a post-write mismatch, or an overflow-assignee that did not resolve to exactly one active User).

before.existed records whether a matching record was present at start (with its prior capacity_weight/routing_model when so); id is populated for every non-blocked status; blocking_issue is non-null only for blocked.

Limitations

  • Attribute-based routing (IsAttributeBased=true) is out of scope — it requires an additional attribute schema.
  • Only MostAvailable and LeastActive are supported on v66.
  • Per-channel capacity weighting is deprecated on v66 (see service-omni-service-channel-configure); per-agent capacity totals are configured elsewhere.

References

FileWhen to read
references/api-notes.mdOn unexpected errors or when customizing fields — QueueRoutingConfig schema, RoutingModel semantics, and how CapacityWeight interacts with agent capacity
scripts/tests/test_queue_routing_config_contracts.pyWhen validating changes — run python3 scripts/tests/test_queue_routing_config_contracts.py from this skill directory

Signals

GitHub stars
1k
Forks
348
Last commit
Sep 2026

ahel review

  • K6low
    bundled executables the agent is told to run

Automated review, not a security audit. Ruleset v1+k2.

Advanced
Item type
skill
Key
service-omni-queue-routing-config-deploy
Source
github.com/forcedotcom/sf-skills