The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes standard git and gh (GitHub CLI) commands to interact with repository refs. All operations that modify the repository, such as branch deletion, require multi-step classification and explicit human confirmation before execution. The skill uses tip SHAs to ensure operations are precise and reversible.- [EXTERNAL_DOWNLOADS]: The skill references a JSON schema located at raw.githubusercontent.com/TitusKirch/skills/main/tituskirch-skills.schema.json. This is a trusted resource managed by the skill's author and is used solely for the purpose of validating configuration files.- [PROMPT_INJECTION]: The skill includes a specific 'Author authority' section that provides a mandatory evidence chain for handling untrusted data. It instructs the agent to treat external strings (e.g., branch names, PR metadata) as data rather than instructions, effectively mitigating indirect prompt injection risks.- [SAFE]: Extensive guardrails are implemented to protect critical repository components. This includes automatic protection for default branches, integration branches, active worktrees, and branches associated with open pull requests. The skill also provides one-line restore commands for every deleted branch to facilitate recovery.

prune-branches