OpenSkillIndex← back to search
claude-code

Sf-skills sf-datacloud-connect

install
source · Clone the upstream repo
git clone https://github.com/Jaganpro/sf-skills
Claude Code · Install into ~/.claude/skills/
T=$(mktemp -d) && git clone --depth=1 https://github.com/Jaganpro/sf-skills "$T" && mkdir -p ~/.claude/skills && cp -r "$T/skills/sf-datacloud-connect" ~/.claude/skills/jaganpro-sf-skills-sf-datacloud-connect && rm -rf "$T"
manifest: skills/sf-datacloud-connect/SKILL.md
source content

sf-datacloud-connect: Data Cloud Connect Phase

Use this skill when the user needs source connection work: connector discovery, connection metadata, connection testing, source-object browsing, connector schema inspection, or connector-specific setup payloads for external sources.

When This Skill Owns the Task

Use 

sf-datacloud-connect
when the work involves:

  • sf data360 connection *
  • connector catalog inspection
  • connection creation, update, test, or delete
  • browsing source objects, fields, databases, or schemas
  • identifying connector types already in use
  • preparing connector definitions for Snowflake, SharePoint Unstructured, or Ingestion API sources

Delegate elsewhere when the user is:

Required Context to Gather First

Ask for or infer:

  • target org alias
  • connector type or source system
  • whether the user wants inspection only or live mutation
  • connection name or ID if one already exists
  • whether credentials are already configured outside the CLI
  • whether the user also expects stream creation right after connection setup
  • whether the source is a database, an unstructured document source, or an Ingestion API feed

Core Operating Rules

  • Verify the plugin runtime first; see ../sf-datacloud/references/plugin-setup.md.
  • Run the shared readiness classifier before mutating connections: 
    node ~/.claude/skills/sf-datacloud/scripts/diagnose-org.mjs -o <org> --phase connect --json
    .
  • Prefer read-only discovery before connection creation.
  • Suppress linked-plugin warning noise with 
    2>/dev/null
    for standard usage.
  • Remember that 
    connection list
    requires 
    --connector-type
    .
  • For 
    connection test
    , pass 
    --connector-type
    when resolving a non-Salesforce connection by name.
  • Discover existing connector types from streams first when the org is unfamiliar.
  • Use curated example payloads before inventing connector-specific credentials or parameters.
  • For connector types outside the curated examples, inspect a known-good UI-created connection via REST before building JSON.
  • Do not promise API-based stream creation for every connector type just because connection creation succeeds.

Recommended Workflow

1. Classify readiness for connect work

node ~/.claude/skills/sf-datacloud/scripts/diagnose-org.mjs -o <org> --phase connect --json

2. Discover connector types

sf data360 connection connector-list -o <org> 2>/dev/null
sf data360 data-stream list -o <org> 2>/dev/null

3. Inspect connections by type

sf data360 connection list -o <org> --connector-type SalesforceDotCom 2>/dev/null
sf data360 connection list -o <org> --connector-type REDSHIFT 2>/dev/null
sf data360 connection list -o <org> --connector-type SNOWFLAKE 2>/dev/null

4. Inspect a specific connection or uploaded schema

sf data360 connection get -o <org> --name <connection> 2>/dev/null
sf data360 connection objects -o <org> --name <connection> 2>/dev/null
sf data360 connection fields -o <org> --name <connection> 2>/dev/null
sf data360 connection schema-get -o <org> --name <connection-id> 2>/dev/null

5. Test or create only after discovery

sf data360 connection test -o <org> --name <connection> --connector-type <type> 2>/dev/null
sf data360 connection create -o <org> -f connection.json 2>/dev/null

6. Start from curated example payloads for external connectors

Use the phase-owned examples before inventing a payload from scratch:

  • examples/connections/heroku-postgres.json
  • examples/connections/redshift.json
  • examples/connections/sharepoint-unstructured.json
  • examples/connections/snowflake-connection.json
  • examples/connections/ingest-api-connection.json
  • examples/connections/ingest-api-schema.json

Typical Ingestion API setup flow:

sf data360 connection create -o <org> -f examples/connections/ingest-api-connection.json 2>/dev/null
sf data360 connection schema-upsert -o <org> --name <connector-id> -f examples/connections/ingest-api-schema.json 2>/dev/null
sf data360 connection schema-get -o <org> --name <connector-id> 2>/dev/null

7. Discover payload fields for unknown connector types

Create one in the UI, then inspect it directly:

sf api request rest "/services/data/v66.0/ssot/connections/<id>" -o <org>

High-Signal Gotchas

  • connection list
    has no true global "list all" mode; query by connector type.
  • The connector catalog name and connection connector type are not always the same label.
  • connection test
    may need 
    --connector-type
    for name resolution when the source is not a default Salesforce connector.
  • An empty connection list usually means "enabled but not configured yet", not "feature disabled".
  • Heroku Postgres, Redshift, Snowflake, SharePoint Unstructured, and Ingestion API all use different credential and parameter shapes; reuse the curated examples instead of guessing.
  • SharePoint Unstructured uses 
    clientId
    , 
    clientSecret
    , and 
    tokenEndpoint
    in the 
    credentials
    array and does not require a 
    parameters
    array.
  • Snowflake uses key-pair auth and can often be created through the API, but downstream stream creation can still remain UI-only.
  • Ingestion API connector setup is incomplete until 
    connection schema-upsert
    has uploaded the object schema.
  • Some external connector credential setup still depends on UI-side configuration or external-system permissions.

Output Format

Connect task: <inspect / create / test / update>
Connector type: <SalesforceDotCom / REDSHIFT / SNOWFLAKE / SPUnstructuredDocument / IngestApi / ...>
Target org: <alias>
Commands: <key commands run>
Verification: <passed / partial / blocked>
Next step: <prepare phase or connector follow-up>

References