The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands via the Bash tool to manage files (ls, cp, mkdir, rm), execute a local validation script (python3 scripts/check-tikz-prevention.py), and invoke document processing tools (xelatex, pdf2svg). These operations are scoped to the local environment and the diagram files being created.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it processes external data.
Ingestion points: TikZ snippets from templates/tikz-snippets/*.tex and user-provided text for diagram edits are read into the agent's context.
Boundary markers: No explicit delimiters or instructions are used to separate the TikZ content when it is passed to the sub-agent reviewer.
Capability inventory: The skill can execute shell commands (Bash), write/edit files, and spawn sub-agents (Task).
Sanitization: There is no evidence of sanitization or filtering of the TikZ content before it is processed or passed to the tikz-reviewer sub-agent.

new-diagram