The Agent Skills Directory

EXTERNAL_DOWNLOADS (MEDIUM): The skill instructions and SKILL.md refer to a canonical repository github.com/tdimino/dabarat which is not a trusted source. It requires the installation or presence of the md_preview_and_annotate package from this external source.
REMOTE_CODE_EXECUTION (MEDIUM): The documentation encourages cloning a remote repository and executing its contents via python3, which bypasses standard package registry verification.
COMMAND_EXECUTION (LOW): The skill utilizes the Bash tool to execute python3 -m commands. While this is the primary purpose of the skill, it relies on the integrity of the external package being executed.
Indirect Prompt Injection (LOW):
Ingestion points: Processes user-supplied .md files for preview and annotation.
Boundary markers: None. The skill does not explicitly warn the agent to ignore instructions embedded within the Markdown files it processes.
Capability inventory: Has permissions to execute python3 commands and write to the filesystem (e.g., ~/.claude/bookmarks/).
Sanitization: The skill's rendering logic for untrusted Markdown content is encapsulated in the external package and cannot be verified.

mdpreview