The Agent Skills Directory

[COMMAND_EXECUTION]: The skill generates a POSIX shell script at ~/.claude/statusline-command.sh and uses chmod +x to make it executable. The script is intended to be run by the Claude Code platform on every update to render a customized status line.
[PROMPT_INJECTION]: The skill handles an indirect prompt injection surface because the generated script ingests JSON data containing environment-derived strings like Git branch names and worktree names. Evidence Chain: 1) Ingestion points: JSON data piped to the generated script's stdin. 2) Boundary markers: Absent for the JSON fields. 3) Capability inventory: Shell script execution with tools like jq, git, awk, and printf. 4) Sanitization: The skill includes specific rules to strip control characters and truncate free-text fields to 30 characters to prevent command or terminal injection.

statusline-config