The Agent Skills Directory

[PROMPT_INJECTION] (SAFE): The skill uses instructional language consistent with technical documentation. No patterns of 'ignore previous instructions', 'bypass', or 'DAN' were detected.
[DATA_EXFILTRATION] (SAFE): No hardcoded credentials (API keys, tokens) or unauthorized network calls were found. The code snippets correctly demonstrate using ctx.auth for identity verification.
[EXTERNAL_DOWNLOADS] (SAFE): The skill references standard convex libraries. An example using fetch to an external API is provided within the context of a Convex 'action', which is the platform's standard way to handle side effects, and is not considered a security risk in this instructional context.
[REMOTE_CODE_EXECUTION] (SAFE): There are no patterns involving piping remote scripts to shells or the use of unsafe execution functions like eval() or exec().

convex