The Agent Skills Directory

[PROMPT_INJECTION]: The instructions do not contain any markers or patterns intended to bypass agent safety guidelines, override system behavior, or extract system prompts.
[COMMAND_EXECUTION]: Uses standard Unix utilities like cat and rg (ripgrep) to inspect local repository files for configuration and form code. These operations are restricted to the local environment and are necessary for the skill's stated purpose of enhancing forms.
[DATA_EXFILTRATION]: No sensitive file paths (e.g., .ssh, .env, .aws) are accessed. There are no network operations, such as curl or wget, targeting non-whitelisted or suspicious domains.
[REMOTE_CODE_EXECUTION]: There is no evidence of downloading remote scripts for execution or installing unverified packages. The execution patterns focus on local build and verification tools.
[EXTERNAL_DOWNLOADS]: Mentions the use of Firecrawl for researching documentation. This is documented as a standard research step for fetching current library guidance and does not involve executing untrusted code.
[SAFE]: The skill operates within the expected bounds of a development tool, focusing on code quality and accessibility standards without introducing security vulnerabilities.

enhance-web-forms