The Agent Skills Directory

[DYNAMIC_EXECUTION]: The verification script scripts/verify-lsp.ts utilizes dynamic module loading via await import() on paths computed at runtime through the findUp utility. This technique is used to load platform-specific engine components.
[REMOTE_CODE_EXECUTION]: The skill's documentation suggests installing system tools using piped shell scripts from remote URLs, such as Bun (https://bun.sh/install | bash) and Julia (https://install.julialang.org | sh). These sources are recognized as well-known technology services.
[COMMAND_EXECUTION]: The skill facilitates the execution of numerous system-level commands, including package managers like npm, pip, go, gem, and dotnet for toolchain setup.
[INDIRECT_PROMPT_INJECTION]: The skill scans project directories and parses JSON configuration files (.codex/lsp-client.json, .opencode/lsp.json), creating an ingestion surface for potentially malicious project data. Ingestion points: scripts/detect-lsp.ts reads filesystem metadata and parses configuration files. Boundary markers: Absent. Capability inventory: System-wide package installation and execution of LSP binaries. Sanitization: Absent; configuration is processed via standard JSON parsing.

lsp-setup