The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes standard local shell commands including git, gh (GitHub CLI), and du to perform repository initialization, commits, and releases. These operations are within the expected scope of a developer tool.
[DATA_EXFILTRATION]: The skill implements a proactive scanning mechanism to identify potentially sensitive files such as .env, tokens, and private keys. It prompts the user to add these to .gitignore before pushing to GitHub, which serves as a security feature to prevent accidental data exposure.
[PROMPT_INJECTION]: The skill processes untrusted data from the local file system, such as filenames and project structures, to generate commit messages and release notes. This represents a surface for indirect prompt injection (Category 8). The mandatory evidence chain includes: (1) Ingestion points: local file system and git status output; (2) Boundary markers: absent; (3) Capability inventory: shell command execution via git and gh; (4) Sanitization: absent. The risk is mitigated by the design requirement for manual user confirmation before critical actions.

git-push