The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The documentation frequently instructs the user or agent to run bunx create-tui@latest, which fetches and executes a package from a remote registry. This pattern is found in SKILL.md, references/core/REFERENCE.md, references/react/REFERENCE.md, and references/solid/REFERENCE.md.
[REMOTE_CODE_EXECUTION]: The instructions for using bunx to initialize projects represent a remote code execution vector as the package source (anomalyco/opentui) is not among the verified trusted organizations.
[COMMAND_EXECUTION]: The file references/core/gotchas.md explicitly references the Bun.$ API for shell command execution (e.g., Bun.$ls -la``). Using such APIs to process terminal commands presents a risk if the input is not strictly controlled.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection. Ingestion points: User input components such as <input> and <textarea> defined in references/components/inputs.md. Boundary markers: None mentioned in the documentation. Capability inventory: Shell execution via Bun.$ (referenced in references/core/gotchas.md) and file system operations via Bun.file (referenced in references/core/gotchas.md). Sanitization: No sanitization or validation strategies are provided for handling user-supplied data within the TUI components.

opentui