The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires the agent to include a remote script from https://www.budge.design/budge.iife.js in the application's layout file (e.g., app/layout.tsx). This domain is not recognized as a trusted vendor or well-known service in the current configuration.
[REMOTE_CODE_EXECUTION]: By instructing the agent to embed the third-party script via <Script> or <script> tags across various frameworks (Next.js, Remix, Astro, etc.), the skill facilitates the execution of unverified remote code within the production or development environment of the application being built.
[INDIRECT_PROMPT_INJECTION]: The skill creates a feedback loop where an external script generates content for the user to paste back to the agent.
Ingestion points: Untrusted data enters the agent context when the user pastes the "submitted prompt" copied to the clipboard by the Budge widget (triggered in SKILL.md Step 5).
Boundary markers: None. There are no delimiters or instructions to the agent to treat the pasted content as potentially adversarial data.
Capability inventory: The agent is instructed to perform file modifications using StrReplace based on the content of the pasted prompt.
Sanitization: There is no validation or sanitization of the instruction generated by the external JavaScript before the agent executes the requested changes.

budge