The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions for the agent to use a local Python script (scripts/install-external-skills.py) to install external capabilities such as browserbase-search. This execution is scoped to skill management and dependency resolution within the local environment.
[INDIRECT_PROMPT_INJECTION]: The skill serves as an ingestion point for user requests related to search optimization. It incorporates a security boundary by explicitly directing the agent to prioritize local rules, safety gates, and product constraints over information retrieved from external guidance or skills.
[METADATA_POISONING]: The .skillmeta.yml file correctly defines the origin and personalization policies, ensuring that only specified paths are allowed for upstream contributions, which prevents unauthorized modifications to sensitive configuration files.

discoverability