Demo org audit

Org Audit — Format & Procedure

Save to: orgs/[alias]-[customer]/audit-[YYYY-MM-DD]-[HHmm].md

• alias from sf config get target-org
• customer = lowercase-hyphenated customer name provided by SE during Stage 0 (e.g. makana-medtech, deutsche-fachpflege)
• HHmm = local time at audit creation (e.g. 0930, 1445)

Use MCP retrieve_metadata for metadata and run_soql_query for record counts.

If MCP unavailable: "MCP is not responding. Quit VS Code fully (CMD+Q) and reopen."

Required Content

Standard Objects in Use

For each standard object commonly used in demos (Account, Contact, Opportunity, Case, Lead, and any others present):

• Label and API name
• Record count
• Record types available
• Active page layout per record type — query ProfileLayout via Tooling API:

  SELECT Layout.Name, RecordType.Name
  FROM ProfileLayout
  WHERE TableEnumOrId = '[Object]'
  AND Profile.Name = 'System Administrator'
  

  List the active layout name for each record type. Flag these explicitly — they are the primary build surface.
• Field-count signal (standard objects only): total field count + custom field count via SELECT QualifiedApiName FROM FieldDefinition WHERE EntityDefinition.QualifiedApiName = '[Object]'. Do NOT retrieve layout XML for the field dump — full field lists are fetched on demand at spec time (sparring Stage 5b), scoped to the locked scenario's objects.

Custom Objects

• API name, label, record count
• Active page layout per record type (name only — same ProfileLayout query as above; no layout-XML field dump)

Existing Lightning Apps

• App API name and label
• Tabs included
• Which app is set as default for System Administrator — this is the primary demo app
• Note which standard objects are already tabbed in this app

Existing Flows

• Name, type, active/inactive, trigger object, brief logic summary
• Flag any flows on objects likely to be used in the demo scenario

Existing LWC Components

• Name, purpose if inferrable
• Which page(s) they appear on if determinable

Existing Custom Permission Sets

• Custom only (exclude standard and managed)

Existing Agentforce Agents and Topics

• If any: name, topics, active/inactive status, AgentType
• ★ Authoring mode / upgrade flag — flag any UI-built agent (pre-filter: AgentType='EinsteinServiceAgent' or Type='ExternalCopilot') whose GenAiPlannerBundle does not retrieve via Metadata API. These can't be edited safely as metadata; topic/action iteration triggers the sparring net-new-vs-upgrade fork. AgentType alone is only the pre-filter — it does NOT change on upgrade, so the planner-retrieve probe is the actual trigger. Topics/descriptions are NOT enumerable on a non-retrievable agent. The sub-agent prompt prompts/sparring/audit/apps-flows-agents.md carries the full procedure.

Notable Gaps and Risks

• Objects with no records (data seeding required)
• Managed package components (prefixed) that cannot be modified
• Execution order conflicts from existing active flows
• Any components already marked ⚠️ from a previous session

---

Audit Priority Flag

Mark the following clearly in the audit output with ★:

• The default Lightning app for System Administrator
• The active page layout for each standard object record type in scope
• Any existing custom objects directly relevant to the demo scenario
• Any UI-built Agentforce agent whose planner won't retrieve via Metadata API (pre-filtered by AgentType='EinsteinServiceAgent' / Type='ExternalCopilot') — topic/action iteration triggers the sparring net-new-vs-upgrade fork

These starred items are the primary build surface. Scout Sparring will use them to anchor scenario design before proposing any new metadata.