The Agent Skills Directory

[COMMAND_EXECUTION]: The core functionality of the skill resides in scripts/repo_tidy.py, which invokes numerous Git commands using subprocess.run. These operations include fetching updates from remotes, switching branches, and deleting local branches or worktrees. The use of a list of arguments for subprocess.run follows best practices to avoid shell injection vulnerabilities, but the script inherently performs destructive local file system operations (branch deletion) as part of its management features.
[INDIRECT_PROMPT_INJECTION]: The skill provides scripts/git-repo-status.sh, designed for use as a SessionStart hook. This script extracts information from the current Git environment (e.g., the current branch name and repository folder name) and outputs a status line that is injected into the agent's prompt context. This creates a surface where maliciously named repositories or branches could attempt to influence the agent's behavior via prompt injection.
Ingestion points: The output of scripts/git-repo-status.sh is intended to be injected into the session start metadata.
Boundary markers: The output uses a simple [repo-status] prefix but lacks strict delimiters or instructions for the agent to ignore potentially malicious embedded content within the environment data.
Capability inventory: The skill possesses the ability to execute Git commands, modify local repository state, and perform network fetches from configured Git remotes.
Sanitization: The status hook does not sanitize branch names or directory names before including them in the output line, relying instead on Git's internal character restrictions for branch names.

repo-tidy