The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the bun runtime to execute shell commands, specifically for running the kordoc conversion utility and local post-processing scripts.
[EXTERNAL_DOWNLOADS]: The skill uses bunx to download and run the kordoc package from the npm registry at runtime. It also mentions installing the Bun runtime via a script from bun.sh, which is recognized as a well-known service.
[INDIRECT_PROMPT_INJECTION]: The skill converts untrusted HWP/HWPX documents into Markdown to be used as context for the agent, which is a potential vector for indirect prompt injection.
Ingestion points: Processes user-provided .hwp, .hwpx, and .hwpml files as described in SKILL.md and resources/execution-protocol.md.
Boundary markers: Absent; there are no instructions for the agent to wrap the converted output in delimiters or to ignore instructions embedded within the processed text.
Capability inventory: The skill has file system access (read/write) and can execute arbitrary shell commands via bun and bunx.
Sanitization: Includes link sanitization within the kordoc tool and a script to strip Private Use Area (PUA) characters in resources/flatten-tables.ts.
[REMOTE_CODE_EXECUTION]: The skill executes remote code by invoking bunx kordoc@latest, which dynamically fetches and runs a tool from the npm registry. The documentation also provides instructions for a curl | bash execution to install the Bun runtime from a well-known service.

oma-hwp