The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes multiple git and CLI commands (gh, glab) to manage repository branches and pull requests. These operations are performed using local scripts that properly quote variables to prevent command injection.
[INDIRECT_PROMPT_INJECTION]: The skill processes file paths and directory names from the local repository diff.
Ingestion points: analyze.sh collects file paths via git diff --numstat.
Boundary markers: Findings are presented to the agent in a structured JSON format.
Capability inventory: The agent can execute git branch manipulation and PR management commands.
Sanitization: The script uses jq to escape file paths for JSON output, mitigating risk from malicious filenames.
[SAFE]: No malicious patterns such as credential exfiltration, persistence, or remote code execution were found. The skill demonstrates defensive coding by using the -- separator in git commands to ensure filenames are not treated as flags.

split-branch