The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes several local bash scripts to perform environment discovery and project setup.
The scripts (check-existing.sh, detect-branch.sh, detect-tech-stack.sh, detect-versions.sh, and verify-setup.sh) use standard system utilities like find, git, grep, and sed to analyze the project structure.
These scripts are part of the skill's own package and do not involve remote downloads or external command execution.
[DYNAMIC_CONTEXT_INJECTION]: The skill uses the !command syntax in SKILL.md to inject project-specific information into the agent's context at load time.
This includes running scripts to detect the current git branch and tech stack.
One injection (gitignore-add.sh) performs a side-effect by modifying the .gitignore file. While typically context injections are read-only, this action is directly aligned with the 'init' purpose of the skill and occurs within the project's own directory.
[DATA_EXPOSURE]: The analysis confirms that the skill only accesses standard project configuration files (e.g., Gemfile, package.json, pyproject.toml) to extract version information and framework names.
No access to sensitive directories (like .ssh, .aws) or environment secrets was detected.
No network operations or exfiltration patterns were found.

init