The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill downloads a project template from https://github.com/Eng0AI/nextjs-supabase-template.git. This repository is not from a verified or well-known source.
[REMOTE_CODE_EXECUTION]: The skill instructs the agent to clone the external repository and then execute npm install, npm run build, and npm run dev. This sequence executes code from an unverified third-party repository on the user's system, which is a high-risk operation.
[INDIRECT_PROMPT_INJECTION]: The skill has a vulnerability surface for indirect prompt injection by ingesting and executing untrusted external content.
Ingestion points: Files downloaded via git clone from the remote repository in SKILL.md.
Boundary markers: None. The agent is instructed to directly treat the cloned files as the project source and execute commands within that environment.
Capability inventory: The agent is granted capabilities to execute shell commands via npm and perform file system operations.
Sanitization: There is no evidence of sanitization or validation of the downloaded repository content before the setup commands are executed.

nextjs-supabase