mcpField
S
segmenting-datacloud

segmenting-datacloud

github.com/forcedotcomData & databases1.1k2.5k installsUpdated 1d ago
Repository

About segmenting-datacloud

A skill for Salesforce Data Cloud audience and insight work. It guides users through creating, publishing, inspecting, and troubleshooting segments and calculated insights, with explicit verification using counts or SQL.

Key features

  1. 01Create and inspect segments and calculated insights
  2. 02Publish segments and run calculated insights
  3. 03Verify results with counts or SQL
  4. 04Troubleshoot segment SQL and member counts

Use cases

  1. 01Create and publish an audience segment
  2. 02Run a calculated insight and inspect aggregate results
  3. 03Investigate empty segments or member-count issues

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/forcedotcom/sf-skills --skill segmenting-datacloud

After installing, type/segmenting-datacloudin Claude Code to confirm it loaded.

Bundled filesCREDITS.mdREADME.md

Details

From the project README

segmenting-datacloud: Data Cloud Segment Phase

Use this skill when the user needs audience and insight work: segments, calculated insights, publish workflows, member counts, or troubleshooting Data Cloud segment SQL.

When This Skill Owns the Task

Use segmenting-datacloud when the work involves:

  • sf data360 segment *
  • sf data360 calculated-insight *
  • segment publish workflows
  • member counts and segment troubleshooting
  • calculated insight execution and verification

Delegate elsewhere when the user is:

Required Context to Gather First

Ask for or infer:

  • target org alias
  • unified DMO (Data Model Object) or base entity name
  • whether the user wants create, publish, inspect, or troubleshoot
  • whether the asset is a segment or calculated insight
  • expected success metric: member count, aggregate value, or publish status

Core Operating Rules

  • Treat Data Cloud segment SQL as distinct from CRM SOQL.
  • Run the shared readiness classifier from the orchestrating-datacloud skill before mutating audience assets: node ../orchestrating-datacloud/scripts/diagnose-org.mjs -o <org> --phase segment --json.
  • Prefer reusable JSON definitions for repeatable segment and CI creation.
  • Use --api-version 64.0 when segment creation behavior is unstable on newer defaults.
  • Verify with counts or SQL after publish/run steps instead of assuming success.
  • Use SQL joins rather than segment members when readable member details are needed.

Recommended Workflow

1. Classify readiness for segment work
node ../orchestrating-datacloud/scripts/diagnose-org.mjs -o <org> --phase segment --json
2. Inspect current state
sf data360 segment list -o <org> 2>/dev/null
sf data360 calculated-insight list -o <org> 2>/dev/null
3. Create with reusable JSON definitions
sf data360 segment create -o <org> -f segment.json --api-version 64.0 2>/dev/null
sf data360 calculated-insight create -o <org> -f ci.json 2>/dev/null
4. Publish or run explicitly
sf data360 segment publish -o <org> --name My_Segment 2>/dev/null
sf data360 calculated-insight run -o <org> --name Lifetime_Value 2>/dev/null
5. Verify with counts or SQL
sf data360 segment count -o <org> --name My_Segment 2>/dev/null
sf data360 query sql -o <org> --sql 'SELECT COUNT(*) FROM "UnifiedssotIndividualMain__dlm"' 2>/dev/null

High-Signal Gotchas

  • Segment creation can require --api-version 64.0.
  • segment members returns opaque IDs; use SQL joins when human-readable member details are needed.
  • Segment SQL is not SOQL.
  • Calculated insight assets and segment SQL have different limitations.
  • Publish/run steps may kick off asynchronous work even when the command returns quickly.
  • An empty segment or calculated-insight list usually means the module is reachable but unconfigured, not unavailable.

Output Format

Segment task: <segment / calculated-insight>
Action: <create / publish / inspect / troubleshoot>
Target org: <alias>
Artifacts: <definition files / commands>
Verification: <member count / query result / publish state>
Next step: <act / retrieve / follow-up>

References

Comments

Markdown supported