The Agent Skills Directory

[COMMAND_EXECUTION]: The skill launches the LibreOffice process headlessly via subprocess.Popen in scripts/uno_bridge.py to handle document edits. This is a functional requirement for the skill's purpose but represents an execution surface for local binaries.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by reading data from external ODT and MD files without protective boundaries or sanitization. Ingestion points: Documents are read using read_text in scripts/writer/session.py and during source imports in scripts/writer/core.py. Boundary markers: Content from files is ingested without delimiters or specific instructions to treat it as untrusted. Capability inventory: The skill can execute local commands and perform file system writes. Sanitization: Text extracted from documents is not validated or sanitized before being provided to the agent.

libreoffice-writer