SvelteKit WebApp

Scaffold and configure a production-ready SvelteKit PWA with opinionated defaults. Use when: creating a new web application, setting up a SvelteKit project,...

This skill appears to be a realistic scaffolding and deployment helper for SvelteKit, but it has a few mismatches you should be aware of before installing or granting tools: - The SKILL.md expects local CLIs (pnpm/pnpx sv, gh, vercel) and will run shell commands to create repos, commit, and push — ensure you want an agent to run those commands and that your CLI is authenticated the way you intend. - The skill will read a user config at ~/.openclaw/workspace/SKILL-CONFIG.json (not declared in metadata). If that file contains sensitive data, review it first or remove it. - The 'Ralph' autonomous prompt instructs the agent to append to files, run tests, and commit/push. Confirm you will approve each action (the doc says commands require approval) and that you will review PRD/progress.txt changes before allowing pushes. Recommendations: - Review the full SKILL.md (especially the PRD and Ralph prompt) and verify you are comfortable with an agent creating repos and pushing code. - Keep backups of any repositories the skill might modify and run the scaffold locally yourself first if unsure. - If you want tighter controls, require the skill to declare required tools/paths explicitly or run the scaffolding steps manually following the generated PRD. If you want, I can highlight the exact lines in SKILL.md that reference the home config, the CLI checks, and the autonomous agent prompt so you can review them before proceeding.

Type: OpenClaw Skill Name: sveltekit-webapp Version: 1.1.1 The skill is classified as suspicious due to its explicit requirement for the `exec` tool, which allows arbitrary shell command execution. While the `SKILL.md` and `references/cli-commands.md` files clearly state that all commands require explicit user approval via the agent's safety framework, the inherent risk of shell access means a sophisticated prompt injection could potentially bypass these controls if the agent's input sanitization or approval process is flawed. However, there is no evidence of intentional malicious behavior, data exfiltration, persistence mechanisms, or obfuscation within the provided files; all listed commands are standard for web application development and deployment.