The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local bash script scripts/discover-toolchain.sh and interacts with CLI tools such as git, gh, and glab to probe repository topology and authentication status.
[INDIRECT_PROMPT_INJECTION]: The skill scans repository manifest files like package.json, Cargo.toml, and pubspec.yaml to infer the project's technology stack. This represents an attack surface for untrusted repositories, but the skill limits its use to existence checks and string matching for configuration purposes.
[SAFE]: The skill demonstrates strong security posture by implementing a mandatory 'Changeset' authorization gate before making any changes. It also explicitly forbids the inspection of sensitive files like .env, private keys, or SSH credentials, and automatically configures .gitignore to prevent leaking machine-local profiles.

zenforge-init