The Agent Skills Directory

[DATA_EXFILTRATION]: The script scripts/shopify_init.py is designed to search for and read .env files not only in the current directory but also in parent directories (specifically ../.env and ../../.env). This behavior allows the script to access the .claude/.env file, which typically contains sensitive API keys, tokens, and other credentials belonging to the agent's environment.
[COMMAND_EXECUTION]: The script scripts/shopify_init.py executes shell commands using the subprocess.run function. While the current implementation specifically calls shopify version, this represents a capability for local command execution within the agent's host environment.
[INDIRECT_PROMPT_INJECTION]: The interactive scaffolding script scripts/shopify_init.py creates a surface for indirect injection by ingesting untrusted data via user prompts and writing it to configuration files.
Ingestion points: Interactive input() prompts in scripts/shopify_init.py for fields like app_name and scopes.
Boundary markers: None identified; user-provided strings are directly embedded into file templates for shopify.app.toml and package.json.
Capability inventory: The script possesses both file-writing capabilities and the ability to execute system commands via subprocess.run.
Sanitization: There is no significant sanitization or validation performed on the user input beyond basic whitespace stripping.
[METADATA_POISONING]: The README.md file contains a deceptive claim that the project is "Maintained By: Claude Code Engineering Team," which contradicts the actual author information. This misleading metadata could lead users to incorrectly attribute the skill to official platform developers, potentially increasing the likelihood of executing its scripts without due caution.

shopify