The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) due to its core functionality of reading external files.
Ingestion points: The agent extracts text from untrusted .pptx files using the officecli view text and officecli view annotated commands.
Boundary markers: The instructions do not define delimiters or specific 'ignore' directives to isolate extracted text from the agent's logic.
Capability inventory: The agent has extensive capabilities to write files, modify content, and execute local CLI commands.
Sanitization: There is no evidence that extracted content is sanitized or validated before being processed by the agent.
[REMOTE_CODE_EXECUTION]: The skill provides instructions to download and execute shell scripts from the author's GitHub repository (https://raw.githubusercontent.com/iOfficeAI/OfficeCli/main/install.sh) by piping them directly into bash or iex. While this is a common method for installing vendor-owned CLI tools and targets the skill author's infrastructure, it remains a high-risk execution pattern.
[EXTERNAL_DOWNLOADS]: The skill performs automated network requests to GitHub's API (api.github.com) to check for updates and uses curl to fetch installation scripts. These operations are limited to the vendor's repositories and are used for maintaining the necessary local environment.
[COMMAND_EXECUTION]: The skill relies on executing the officecli binary through various subprocess calls to perform presentation tasks. These commands are the intended primary mechanism of the skill.

officecli-pptx