The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill's setup instructions utilize piped remote execution for installation. On macOS and Linux, it pipes a script from GitHub directly into bash: curl -fsSL https://raw.githubusercontent.com/iOfficeAI/OfficeCLI/main/install.sh | bash. On Windows, it uses Invoke-Expression (iex) to execute a remote PowerShell script: irm https://raw.githubusercontent.com/iOfficeAI/OfficeCLI/main/install.ps1 | iex.
[EXTERNAL_DOWNLOADS]: To function, the skill requires downloading and installing an external CLI utility ('officecli') from the author's GitHub repository.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests untrusted data from external sources.
Ingestion points: File data is read into the agent's context using officecli view and officecli get in SKILL.md and editing.md.
Boundary markers: The instructions lack delimiters or warnings to ignore instructions found within the spreadsheet data.
Capability inventory: The skill has the ability to execute shell commands (officecli) and modify the local file system (creating and editing workbooks).
Sanitization: There is no evidence of sanitization or filtering applied to the content extracted from .xlsx or .csv files before it is processed by the agent.
[COMMAND_EXECUTION]: The skill relies on executing a local CLI tool to perform spreadsheet operations. This includes the raw-set command which allows the agent to inject arbitrary XML into the workbook structure using XPath selectors, providing high-privilege access to the file's internal XML components.

officecli-xlsx