The Agent Skills Directory

[COMMAND_EXECUTION]: The skill performs runtime compilation of C source code into a shared object file using gcc and injects it into the soffice process environment using LD_PRELOAD to handle socket restrictions. Multiple scripts also execute system commands including soffice, pdftoppm, and git via subprocess.run.
[EXTERNAL_DOWNLOADS]: The skill instructions and documentation prompt for the installation of various third-party dependencies from the NPM registry, such as pptxgenjs, react-icons, and sharp.
[PROMPT_INJECTION]: The skill is designed to extract and process text content from user-provided .pptx files, which represents an indirect prompt injection surface where malicious instructions in a presentation could influence the agent's behavior.
Ingestion points: Text and XML extraction from PowerPoint slides as documented in SKILL.md and references/editing.md.
Boundary markers: No specific delimiters or instructions to ignore embedded commands are used when presenting extracted data to the agent.
Capability inventory: The skill has access to significant capabilities including shell command execution (subprocess.run), file system modification (write/delete), and network operations via its dependencies.
Sanitization: Employs defusedxml for XML parsing to mitigate risks like XML External Entity (XXE) attacks.

pptx