The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches component definitions and metadata from https://ui.ondo.dou.so/r/. This domain and its parent have been explicitly flagged as malicious (phishing) by automated security scanners.
[REMOTE_CODE_EXECUTION]: The skill executes remote packages, including @dou.so/ondo-ui and skills, using runners such as bunx, npx, and pnpm dlx. These packages are retrieved from infrastructure associated with phishing alerts, posing a high risk of malicious code injection and execution during the onboarding process.
[COMMAND_EXECUTION]: The skill constructs shell commands using user-provided inputs, such as project names and installation paths, without explicit sanitization. This creates a vulnerability where malicious input could lead to command injection, especially given the skill's instructions to execute all steps autonomously without further user verification.
[CREDENTIALS_UNSAFE]: The skill accesses and modifies the global configuration file at ~/.codex/config.toml. This file typically contains sensitive authentication tokens and settings for MCP clients. Modifying this file outside the project scope can be used as a persistence mechanism or lead to the exposure of credentials.

ondo-onboarding