service-omni-agent-work-sharing-configure
SkillAI & modelsUse to configure or verify org-wide AgentWork visibility for supervisors after explicit confirmation. Triggers: let supervisors read AgentWork, configure AgentWork sharing, verify AgentWork OWD, make AgentWork visible to supervisors, change AgentWork internal access. Do not use to filter which reps appear in Command Center; use service-omni-supervisor-config-deploy for that.
Available today. Use it from your connected AI after setup.
No other account needed.
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-agent-work-sharing-configure skill
What this skill tells your AI
The instructions your AI receives, as published by forcedotcom/sf-skills in skills/service-omni-agent-work-sharing-configure/SKILL.md and read by ahel’s review.
Detect and optionally configure the standard AgentWork object's internal organization-wide default. The supported headless contract is CustomObject:AgentWork metadata: retrieve the current metadata, preserve externalSharingModel, deploy internal sharingModel=Read, and retrieve it again for proof.
This setting grants read visibility broadly to internal users who already have object access. It is not a substitute for an OmniSupervisorConfig: that configuration filters which reps, queues, skills, actions, and tabs a supervisor sees, but does not grant record access.
Inputs
bash scripts/configure-and-report.sh <org-alias> --plan
bash scripts/configure-and-report.sh <org-alias> --confirm-org-wide-visibility
org-aliasis required and must already be authenticated withsf.--planretrieves and reports the current and desired models without deploying.--confirm-org-wide-visibilityis required only when a run must changePrivatetoRead. Omitting it blocks before deployment.
Safety contract
- A run refuses production customer orgs. Plan mode remains read-only.
externalSharingModelis preserved exactly.ReadandReadWritealready provide internal read visibility and are reused. The skill never narrowsReadWritetoRead.- The only supported mutation is
PrivatetoRead; the skill never grants internal edit access. - Failed classification, retrieve, deploy, or read-back blocks. It never reports success from an inconclusive response.
- Use a dedicated sandbox or test org because changing an OWD can initiate sharing recalculation.
Workflow
- Classify the target org and enforce the production-write guard.
- Retrieve
CustomObject:AgentWorkthrough Metadata API. - Read internal and external sharing models.
- Reuse
ReadorReadWritewith no deploy. - In plan mode, report the proposed
PrivatetoReadchange and exit yellow. - In run mode, require explicit blast-radius confirmation.
- Deploy only the standard-object metadata containing preserved external sharing and internal
Read. - Retrieve again and require internal
Readplus the unchanged external model.
Output
A single JSON object containing status, previous_sharing_model, sharing_model, external_sharing_model, changed, dry_run, deploy_id, and blocking_issue.
configured: a confirmedPrivatetoReaddeployment succeeded and read-back matched.reused: currentReadorReadWritealready provides visibility.action_needed: plan mode foundPrivate; no write occurred.blocked: a safety guard or supported-API verification failed.
Limitations
- This skill changes visibility, not supervisor scope. Use the supervisor configuration skill for filtering.
- It does not create sharing rules or per-record
AgentWorkSharerows. - It cannot prove a particular supervisor's complete UI access; validate the end-to-end Command Center journey with the intended supervisor user after configuration.
References
| File | When to read |
|---|---|
references/api-notes.md | Before approving the org-wide change or diagnosing a retrieve/deploy failure |
Signals
- GitHub stars
- 1k
- Forks
- 348
- Last commit
- Sep 2026
Advanced
- Item type
- skill
- Key
service-omni-agent-work-sharing-configure- Source
- github.com/forcedotcom/sf-skills