The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's auditing instructions involve scanning the project codebase for color literals, creating a surface where untrusted data enters the agent's context. \n- Ingestion points: The audit process in palette-structure.md instructs the agent to search for hex, rgb, and oklch literals in project files. \n- Boundary markers: No specific delimiters are used to separate color data from potentially malicious instructions. \n- Capability inventory: The skill requires the ability to read and search files. \n- Sanitization: No sanitization is specified for the data found during the audit. \n- [SAFE]: The skill recommends the use of well-known color manipulation libraries such as culori, colorjs.io, and chroma.js, which are standard industry tools.

better-colors