The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands to discover and delete Git branches and worktrees. While these are destructive operations, the skill mandates a human-in-the-loop confirmation step to mitigate risk.
[INDIRECT_PROMPT_INJECTION]: The skill parses output from git branch -vv, which includes information tracking remote branches. This presents an attack surface where a maliciously named remote branch could attempt to influence the agent's behavior during the deletion phase.\n
Ingestion points: The scripts/clean-gone script reads tracking information derived from the remote repository.\n
Boundary markers: The skill explicitly requires the use of platform blocking tools (such as AskUserQuestion) to get user confirmation for the full list of branches before proceeding.\n
Capability inventory: The agent has the capability to remove worktrees (git worktree remove --force) and delete local branches (git branch -D) as described in SKILL.md.\n
Sanitization: The discovery script scripts/clean-gone performs basic sanitization by excluding the current branch and the HEAD reference, while ensuring branch names are properly extracted without whitespace using awk and sed.

clean-gone-branches