The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill provides guidance on managing sensitive environment variables such as OPENAI_API_KEY and blockchain PRIVATE_KEY. It recommends using .env.local for secret management, which is a standard security practice for local development.
[REMOTE_CODE_EXECUTION]: The documentation includes commands like bunx create-aixyz-app and bun add, which download and execute packages from the public Bun/NPM registry. These actions are standard for project scaffolding and dependency management within the intended use-case of a developer framework.
[PROMPT_INJECTION]: The framework naturally supports building agents that process user-controlled data. This creates a surface for indirect prompt injection where malicious instructions could be embedded in tool inputs. The skill demonstrates the use of zod for input schema validation, which provides a layer of structural sanitization.
[SAFE]: Analysis of the instructions and code snippets reveals no evidence of malicious obfuscation, unauthorized data exfiltration, or privilege escalation. All network references target either well-known model providers or the vendor's own official infrastructure.

aixyz