The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes Python scripts that execute shell commands to interact with the Typst compiler and detect versions.
Evidence: scripts/search-packages.py uses subprocess.run(['typst', '--version'], ...) to identify the local compiler environment.
Evidence: scripts/validate-examples.py uses subprocess.run(['typst', 'compile', ...], ...) to verify the correctness of code snippets within the documentation.
[DYNAMIC_EXECUTION]: The validation suite dynamically generates and executes Typst source files from documentation content.
Evidence: scripts/validate-examples.py extracts typst code blocks from markdown files, prepends a variable preamble, writes them to temporary .typ files, and executes the Typst compiler on them.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface where it processes data (markdown documentation) that influences command execution, creating a path for indirect injection.
Ingestion points: The scripts/validate-examples.py script reads all .md files in the repository to extract code blocks.
Boundary markers: No explicit boundary markers or 'ignore' instructions are used when processing the extracted code blocks.
Capability inventory: The script uses subprocess.run to execute the Typst compiler on extracted content.
Sanitization: The script implements SKIP_PATTERNS (regex) to filter out potentially dangerous or non-compilable patterns such as relative imports, remote packages, image references, and bibliography calls.

typst