The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The lt dev install command instructs users to use sudo -E HOME="$HOME" caddy trust to install a local CA into the system trust store. This requires administrative privileges and affects system-wide security settings.
[PERSISTENCE]: The lt dev install command establishes persistent background services by creating and bootstrapping a LaunchAgent on macOS (~/Library/LaunchAgents/tech.lenne.lt-dev-caddy.plist) or a systemd-user unit on Linux. This ensures the proxy service remains active across user sessions.
[INDIRECT_PROMPT_INJECTION]: The skill recommends using WebFetch to retrieve canonical documentation from external URLs hosted on GitHub (e.g., https://raw.githubusercontent.com/lenneTech/cli/main/docs/LT-ECOSYSTEM-GUIDE.md). This ingestion of external data without explicit boundary markers or sanitization creates a potential surface for indirect prompt injection.
Ingestion points: Documentation URLs listed in the 'External Documentation' section of SKILL.md.
Boundary markers: Absent; the fetched content is intended for deep context integration.
Capability inventory: The skill uses the lt CLI which can modify the file system, execute shell commands, and initiate network tunnels.
Sanitization: No sanitization or validation of the fetched remote content is specified.
[COMMAND_EXECUTION]: The lt git reset command is described as destructive and irreversible, executing git reset --hard followed by a force-pull. This leads to immediate and permanent loss of local work if executed without caution.
[EXTERNAL_DOWNLOADS]: The skill documents commands that clone several repositories from GitHub (lenneTech/cli, lenneTech/nest-base, lenneTech/nest-server-starter, lenneTech/nuxt-base-starter). These are resources associated with the skill's authoring organization and are used for project scaffolding.

using-lt-cli