The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell scripts and the Salesforce CLI (sf) to manage project metadata and discover local components.\n
The sf template generate flexipage command is used to bootstrap valid metadata files.\n
The sf plugins install templates command ensures required CLI functionality is present.\n
Local scripts scripts/scan-lwc-components.sh and scripts/resolve-component-instructions.sh are used to scan the workspace and resolve configuration without network overhead.\n- [PROMPT_INJECTION]: The skill processes untrusted data, including user intent and external component metadata, which represents a surface for indirect prompt injection.\n
Ingestion points: Data enters the agent context via user utterances, local Lightning Web Component (LWC) source files scanned by scripts, and metadata retrieved through the DISCOVER_UI_COMPONENTS and GET_UI_COMPONENT_SCHEMAS MCP actions.\n
Boundary markers: Explicit delimiters for separating untrusted data from instructions are not defined in the skill's prompting logic.\n
Capability inventory: The skill has the capability to execute shell commands (Bash), read files (Read), and write metadata files (Write).\n
Sanitization: The skill provides clear guidance in references/xml_rules.md for entity-encoding HTML content to prevent XML structure corruption.

platform-flexipage-generate