The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill provides instructions to install the officecli tool by piping scripts from https://d.officecli.ai/install.sh and https://d.officecli.ai/install.ps1 directly into bash and iex. This pattern allows remote code to execute with the user's privileges without an opportunity for prior manual review of the script content.
[EXTERNAL_DOWNLOADS]: The skill contains references to downloading binary releases from the vendor's public GitHub repository at github.com/iOfficeAI/OfficeCLI/releases.
[COMMAND_EXECUTION]: The skill uses a variety of shell commands, including curl, jq, grep, and python3, to perform environment setup and document manipulation. It includes an inline Python script that parses XML data extracted from Word documents.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection attack surface as it processes external Word documents that may contain untrusted data.
Ingestion points: External Word documents processed by the officecli tool and shell variables derived from document content.
Boundary markers: None present to instruct the agent to ignore potentially malicious content within the documents.
Capability inventory: File system modification (officecli), network requests (curl), and shell/script execution (bash, python3).
Sanitization: Employs structured parsing with jq and xml.etree.ElementTree, which mitigates structural attacks but does not prevent content-driven logic manipulation.

officecli-word-form