The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses mcp__headless-360__dispatch and mcp__headless-360__dispatch_readonly tools to interact with Salesforce REST APIs. These are legitimate administrative operations for configuring feature flags and assigning permission sets within a managed environment.- [CREDENTIALS_SAFE]: The skill does not handle or store raw credentials, API keys, or secrets. Authentication is managed through OAuth JWTs bound to the MCP session, ensuring that the AI agent never has direct access to sensitive credentials.- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied usernames for permission assignment. It mitigates injection risks by using structured query parameters for SOQL queries and validating that the target user resolves to exactly one record before performing write operations.- [DATA_EXPOSURE]: The skill reads organizational metadata (feature status, permission set IDs) and user information (usernames, IDs). All data access is scoped to the current authenticated Salesforce session and follows the principle of least privilege required for CMDB setup.- [SAFE]: All external operations are directed to forcedotcom (Salesforce) infrastructure, which is the verified vendor for this skill. There are no downloads from untrusted third-party sources or execution of remote scripts from unknown domains.

service-itsm-agentic-setup-cmdb-discovery-configure