The Agent Skills Directory

[PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection because it instructs the agent to process external, potentially untrusted project data.
Ingestion points: The agent is directed to read package.json, tailwind.config.js, metro.config.js, babel.config.js, and project source files (.ts, .tsx, etc.) from the user's workspace.
Boundary markers: There are no explicit instructions or delimiters used to separate the project data from the agent's instructions, nor are there commands to ignore instructions embedded within those files.
Capability inventory: The agent has the authority to execute shell commands (npm, yarn, bun, rg, rm) and modify files, which could be exploited if malicious instructions are present in the ingested data.
Sanitization: No sanitization or validation of the content of the read files is performed before the agent acts on the migration steps.
[COMMAND_EXECUTION]: The skill provides various shell commands for package uninstallation, installation, and codebase searching. These commands are standard for the described migration workflow.
[SAFE]: All external documentation and package references point to the vendor's official domain (uniwind.dev) or standard registries, which is consistent with the skill author (uni-stack).
[SAFE]: The use of instructional markers such as 'CRITICAL' and 'IMPORTANT' is used to emphasize technical migration requirements and common developer errors, rather than attempting to bypass safety filters.

migrate-nativewind-to-uniwind