The Agent Skills Directory

[COMMAND_EXECUTION]: Multiple scripts within the skill, including fill_inspect.py, fill_write.py, merge.py, reformat_parse.py, render_body.py, make.sh, and render_cover.js, utilize subprocess.check_call or spawnSync. These functions are used to execute shell commands to modify the system environment by installing dependencies.
[EXTERNAL_DOWNLOADS]: The skill performs several automated downloads from external services. It retrieves font stylesheets from Google Fonts and uses package managers to download pypdf, reportlab, matplotlib, and playwright from PyPI and NPM. Additionally, it downloads Chromium browser binaries via the Playwright install utility.
[REMOTE_CODE_EXECUTION]: The skill features an automatic dependency installation mechanism that downloads and installs third-party code from public registries at runtime without user confirmation. This occurs during the initialization of several Python scripts and the execution of the Node.js cover renderer, which could lead to the execution of malicious code if a package in the supply chain were compromised.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes untrusted data from Markdown, TXT, and PDF files. Ingestion points: External document content is parsed in reformat_parse.py and rendered in render_body.py. Boundary markers: There are no markers or instructions used to isolate processed data from the agent's instructions. Capability inventory: The skill has the ability to execute shell commands and perform file operations. Sanitization: The skill lacks comprehensive sanitization of input text, relying on basic regular expressions for Markdown-to-XML conversion before rendering content into the PDF.

minimax-pdf