The Agent Skills Directory

[SAFE]: The skill focuses on enhancing code editing safety through LSP integration. It uses speculative previews via the mcp__lsp__preview_edit tool to assess the impact of changes on the code's diagnostic state (errors and warnings) before writing to the filesystem.
[SAFE]: External dependencies are limited to the vendor's own agent-lsp MCP server repository on GitHub, which is necessary for the skill's stated purpose and is a legitimate vendor resource.
[SAFE]: The workflow incorporates explicit user confirmation steps when errors are introduced during the preview phase, preventing the automated commitment of potentially broken code.
[SAFE]: Permission configuration in the YAML frontmatter (tool_permissions) demonstrates a best-practice approach by restricting access to tools like Write or Edit during the setup and speculative preview phases.

lsp-safe-edit