The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to review and modify Android/KMP source code. This creates a surface for indirect prompt injection if the user-provided code contains malicious instructions intended to influence the agent's behavior.
Ingestion points: The skill processes user-supplied Kotlin/Compose source code files during review and editing tasks.
Boundary markers: There are no explicit instructions for the agent to treat user-provided code as untrusted or to use specific delimiters to separate code from instructions.
Capability inventory: The agent is authorized to search the codebase using rg and is expected to generate or modify code based on the provided guidelines.
Sanitization: No explicit sanitization or filtering of the input code content is performed.
[SAFE]: All external URLs referenced in the skill point to official documentation from trusted organizations, including Google (Android/Material Design) and JetBrains (Kotlin/Compose Multiplatform).
Evidence:
https://m3.material.io/styles/color/roles (Google/Material Design)
https://developer.android.com/develop/ui/compose/performance/stability/strongskipping (Google)
https://kotlinlang.org/docs/whatsnew2020.html (JetBrains)
[SAFE]: The skill uses standard command-line tools like rg (ripgrep) solely for the purpose of auditing code quality and ensuring compliance with the defined architectural rules (e.g., checking for hardcoded colors or ViewModel usage).

compose-ui