The Agent Skills Directory

[SAFE]: The skill provides documentation and implementation guidelines for the Nitro server toolkit. It does not contain executable malicious code or instructions that compromise agent safety.\n- [SAFE]: Credential management examples use non-functional placeholders such as "dev_token", "secret", or generic connection strings, avoiding the exposure of actual secrets or sensitive data.\n- [SAFE]: External links and package references are limited to official documentation and repositories for Nitro and its associated libraries (H3, unstorage, db0, etc.), which are well-known and trusted projects in the Node.js ecosystem.\n- [SAFE]: The instructions reinforce secure development practices, specifically advising on the use of parameterized SQL queries via db.sql tagged templates to mitigate injection vulnerabilities and stressing the importance of authentication for runtime tasks and WebSocket upgrades.\n- [INDIRECT_PROMPT_INJECTION]: The skill defines patterns for ingesting untrusted data through HTTP request handlers, WebSocket messages, and task payloads. It outlines structural boundaries through standard handler definitions and provides access to capabilities including file storage, network fetching, and SQL database execution. Sanitization and security are addressed by recommending parameterized queries and manual authentication checkpoints.

nitro