The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute several local Python scripts and a CLI tool during the verification process. The source code for these scripts is not included in the skill bundle, preventing verification of their behavior:
python3 scripts/check_android_external_skill_manifest.py (referenced in SKILL.md)
python3 scripts/workflow.py route ... (referenced in SKILL.md)
vibeguard audit . --rules . (referenced in SKILL.md)
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process and 'distill' rules from external Android skill repositories. This creates a potential vector where malicious instructions in the external source material could influence the agent's logic or guidance:
Ingestion points: The agent is directed to read and apply guidance from ../android-external-skill-source-coverage/, github.com/android/skills, github.com/skydoves/compose-performance-skills, and github.com/chrisbanes/skills.
Boundary markers: There are no explicit instructions or delimiters provided to prevent the agent from obeying instructions found within the external source material.
Capability inventory: The skill bundle references command execution capabilities (Python scripts) that could be triggered or influenced by the processed data.
Sanitization: No evidence of sanitization or validation of the external content is present in the provided instructions.
[EXTERNAL_DOWNLOADS]: The skill references several external GitHub repositories for documentation and guidance. While these are well-known sources in the Android community, they represent external dependencies for the skill's logic:
https://github.com/android/skills (Official Android/Google repository)
https://github.com/skydoves/compose-performance-skills (Community resource)
https://github.com/chrisbanes/skills (Community resource)

source-coverage