The Agent Skills Directory

[SAFE]: The skill implements a robust decision-making framework for developer assistance without executing arbitrary code or accessing sensitive local data outside of standard project inspection bounds.
[SAFE]: External dependencies are restricted to official DevExpress documentation via MCP tools (devexpress_docs_search, devexpress_docs_get_content). The skill explicitly forbids the main agent from calling these tools directly, enforcing a delegation pattern to sub-agents (doc-fetch, doc-search) for better context isolation.
[SAFE]: Hardcoded code snippets (such as the SignalR pattern in source-selection.md) are standard implementation examples for the vendor's own libraries (DevExpress and Microsoft SignalR).
[SAFE]: The inspect sub-agent is limited to a one-hop dependency walk, preventing excessive file system traversal.
[SAFE]: Instructions include clear security-minded constraints, such as treating fetched documentation as data/reference material rather than directives, and advising against running embedded instructions found in remote content.

devexpress-wpf-grid-datasource