The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches content from remote GitHub repositories defined in config.yaml. Sources include the author's own repositories and the official Anthropics skills repository.
[REMOTE_CODE_EXECUTION]: Content downloaded from remote sources is installed into system-level directories (~/.codex/skills and ~/.claude/skills), where it is subsequently loaded and executed by the AI agent.
[COMMAND_EXECUTION]: The scripts/install.py script uses subprocess.run to execute git for cloning repositories and python3 for running installation and cleanup tasks. It also mentions using the gh CLI for bug reporting.
[PROMPT_INJECTION]: The skill represents an indirect prompt injection surface as it is designed to import external instructions (skills) into the agent's operating environment.
Ingestion points: Data is ingested via git clone from remote URLs specified in config.yaml and SKILL.md.
Boundary markers: No specific boundary markers or sanitization are applied to the instructions within the downloaded skills.
Capability inventory: The skill has capabilities for shell command execution (subprocess.run), file system modification (shutil.copytree, shutil.rmtree), and network access via git (in scripts/install.py).
Sanitization: The skill uses MD5 hashing for version control but does not perform content validation or sanitization of the installed instructions.

install-bensz-skills