The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses node to execute a local verification script (scripts/verify-child-verdict.mjs). This is a benign use of command execution to deterministically check the success status of child skills before proceeding with the setup queue. It does not ingest untrusted user input into the shell.
[REMOTE_CODE_EXECUTION]: The skill uses allowed-tools for Bash and node, but it only executes its own distributed script to verify internal state. No remote scripts or unknown packages are downloaded or executed.
[PRIVILEGE_ESCALATION]: The skill performs administrative actions within a Salesforce org (feature enablement, agent activation). However, it includes mandatory safety behavior requiring the agent to confirm the target org with the user and explicitly state that the org will be modified before any writes occur.
[INDIRECT_PROMPT_INJECTION]: The skill processes conversation history to track progress. While this is an ingestion surface, the skill's capabilities are limited to delegating to specific, hard-coded child skills. It does not perform sensitive file operations or network exfiltration that could be easily exploited by data processed from chat history.

service-itsm-agentic-setup-agentforce-coordinate