The Agent Skills Directory

[SAFE]: The skill includes robust safety constraints, such as prohibiting wildcard deployments and requiring explicit target org resolution before execution to prevent errors.\n- [EXTERNAL_DOWNLOADS]: Installation workflows rely on cloning the official repository from GitHub, which is a standard and transparent distribution method for agent skills.\n- [COMMAND_EXECUTION]: Shell commands are strictly limited to the Salesforce CLI (sf), standard filesystem utilities (mkdir, rm), and the Prettier formatter. These tools are used appropriately and safely for their intended developer workflow.\n- [PROMPT_INJECTION]: The skill has a surface for indirect prompt injection through its interaction with external Salesforce orgs.\n
Ingestion points: Metadata names and component content retrieved via sf org list metadata and sf project retrieve start.\n
Boundary markers: Instructions mandate the use of the --json flag to provide structured output for agent parsing.\n
Capability inventory: The skill utilizes subprocess calls to the sf CLI, performs local file writes, and runs npx prettier.\n
Sanitization: No explicit sanitization or filtering of the content retrieved from the org is performed before agent processing.

salesforce-development