The Agent Skills Directory

[METADATA_POISONING]: The skill includes a file (skill-report.json) that contains pre-emptive security audit results, including a 'safe' verdict and claims that all findings are false positives. This is a self-referential attempt to influence automated analyzers and human reviewers, particularly as the author cited in the report ('wshobson') differs from the documented skill author ('aiskillstore').\n- [INDIRECT_PROMPT_INJECTION]: The skill provides UI implementation patterns in references/compose-components.md and SKILL.md (e.g., ItemList, SearchableScreen) that demonstrate how to process and display external data fields like item.name and item.description. These patterns do not include instructions for sanitization or boundary marking, creating a potential surface for indirect prompt injection if the ingested data contains malicious instructions targeting the agent.\n- [EXTERNAL_DOWNLOADS]: The skill documentation correctly references official Android and Material Design resources from Google and Android (e.g., m3.material.io, developer.android.com). These references target well-known and trusted technology services and are documented neutrally for educational purposes.

mobile-android-design