The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes the sf (Salesforce CLI) tool and a local diagnostic script to interact with Salesforce environments.
Evidence: SKILL.md specifies execution of sf data360 commands and node ~/.claude/skills/sf-datacloud/scripts/diagnose-org.mjs.
[DATA_EXFILTRATION]: The skill instructions and templates involve handling sensitive credentials for database and cloud service connections.
Evidence: JSON templates in examples/connections/ manage fields for passwords and secrets (e.g., <HEROKU_DB_PASSWORD>, <AZURE_APP_CLIENT_SECRET>). While the skill processes these secrets, no unauthorized external transmission was identified.
[PROMPT_INJECTION]: The skill is exposed to indirect prompt injection risks by ingesting and acting upon metadata retrieved from external Salesforce organizations.
Ingestion points: sf data360 connection objects and sf data360 connection fields commands in SKILL.md.
Boundary markers: No explicit boundary markers or isolation instructions for external data processing were found.
Capability inventory: The skill has the capability to create/mutate connections and execute arbitrary REST API requests via sf api request rest.
Sanitization: No sanitization logic for fetched metadata is specified in the instructions.

sf-datacloud-connect