The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs users to install the official @vercel/kv package from the NPM registry and pull configuration from Vercel's infrastructure using the official Vercel CLI. These are well-known and trusted sources.
[CREDENTIALS_UNSAFE]: No hardcoded secrets were found. The skill correctly identifies .env.local as a sensitive file and provides clear instructions to add it to .gitignore to prevent accidental exposure of KV tokens.
[COMMAND_EXECUTION]: Code examples and instructions include standard development commands like npm install and vercel env pull. These commands are expected for the skill's purpose and do not represent malicious or hidden execution patterns.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data via key-value storage for sessions and caching. It mitigates injection risks by recommending structured data handling, JSON serialization, and runtime type validation using tools like Zod.
[PRIVILEGE_ESCALATION]: No commands attempting to escalate system privileges, such as sudo or unauthorized file permission changes, were detected.

vercel-kv