The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses find, grep, and gh (GitHub CLI) to search for files and repositories. These are standard utility commands for searching local file systems and public code repositories.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to search and ingest external SKILL.md files from GitHub and the web. This creates a surface for indirect prompt injection if those files contain malicious instructions. However, the skill explicitly mitigates this in 'Step 4
Vet External Matches' by instructing the agent to 'Read the SKILL.md frontmatter and instructions' and 'Look for unexpected shell commands, file writes, network calls, credential handling, or package installs' before recommending or adopting them.
[SAFE]: The local search paths (~/.claude/skills) are standard directories for agent skills, and the use of gh search targets public repositories. No sensitive file access or unauthorized network operations were detected.

skill-scout