The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes claude plugin validate and a local repository script scripts/fetch-claude-docs.sh. These are documented maintenance tasks for plugin contributors and do not involve suspicious parameters or escalated privileges.
[EXTERNAL_DOWNLOADS]: The skill references a mechanism for updating local documentation caches. This is a legitimate functional requirement for identifying schema drift and does not involve downloading untrusted executable code.
[PROMPT_INJECTION]: The skill processes documentation data which serves as a potential surface for indirect prompt injection. However, the skill implements risk mitigation by instructing the agent to use targeted snippets rather than full-page ingestion and delegates the complex interpretation to a specialized sub-agent.
[DATA_EXFILTRATION]: No exfiltration patterns were detected. The skill reads from specific local documentation caches and plugin directories intended for validation purposes without transmitting data to external third-party domains.

docs-check