installing-skills

Pass

Audited by Gen Agent Trust Hub on Oct 13, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill fetches repository structure and skill definition files from the author's GitHub repository (github.com/oaustegard/claude-skills). It connects to api.github.com and raw.githubusercontent.com to retrieve this content.
  • [COMMAND_EXECUTION]: A bash script (install.sh) is used to automate the installation process, performing directory creation with mkdir -p and file retrieval with curl to the local filesystem at /mnt/skills/user.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest external instructions which the agent then integrates into its operational context.
  • Ingestion points: The scripts/install.sh script downloads SKILL.md files from a remote source.
  • Boundary markers: No specific boundary markers or instruction-ignore wrappers are applied to the downloaded content during the installation phase.
  • Capability inventory: The skill utilizes network access (curl), directory management (mkdir), and shell script execution (bash).
  • Sanitization: The script limits downloads to a specific hardcoded repository path associated with the author.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 13, 2026, 06:44 PM