The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill acts as a proxy that ingests untrusted user input from $ARGUMENTS and attachment paths, then interpolates this data into prompt files for execution by secondary agents.
Ingestion points: User-provided task descriptions and file attachment paths (including absolute paths) are received via $ARGUMENTS in SKILL.md.
Boundary markers: Absent. The skill does not use specific delimiters or instructions to prevent the secondary agent from following instructions embedded within the user-provided text.
Capability inventory: The skill has access to the Bash tool for executing shell commands and the Write tool for file creation.
Sanitization: No explicit validation or sanitization is performed on the user-provided content or file paths before they are written to implementation prompts.
[DYNAMIC_EXECUTION]: The skill dynamically generates shell commands and markdown files at runtime. It uses a template pattern (cat > "$tmpfile" << 'EOF') to assemble executable instructions for the agent's environment and subsequent sub-agents.
[COMMAND_EXECUTION]: The skill performs local system operations by executing the workmux command-line utility via the Bash tool to manage git worktrees and tmux sessions.

worktree