The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill uses npx -y bmmd to download and execute an external package at runtime. This practice introduces risk as the code is fetched from a remote registry and executed without version pinning or integrity verification.
[EXTERNAL_DOWNLOADS]: The skill relies on npx to fetch the bmmd package from the NPM registry during execution if it is not already present locally.
[COMMAND_EXECUTION]: The skill instructions direct the agent to execute several shell commands, including node --version, npx -y bmmd, curl, and piping operations (cat | npx).
[DATA_EXFILTRATION]: User-provided Markdown and HTML content is sent to an external REST API at https://bm.md/api/markdown/* for processing. While this is the stated purpose of the skill, it involves transmitting potentially sensitive document data to a third-party service.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted Markdown and HTML data from files or user input and passes it to both CLI tools and remote APIs.
Ingestion points: The skill reads article.md and page.html files and accepts string input for the markdown and html fields.
Boundary markers: There are no explicit instructions or delimiters to prevent the agent from following instructions embedded within the processed content.
Capability inventory: The skill can execute shell commands (npx, curl), write to the filesystem (--output), and perform network operations.
Sanitization: No sanitization or validation of the processed content is specified before execution or transmission.

bm-md