The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the markit CLI for document and media conversion. It also includes an onboard command that modifies local instruction files (e.g., CLAUDE.md) to incorporate tool usage guidelines, which functions as a form of persistent behavioral modification for the agent environment.
[EXTERNAL_DOWNLOADS]: The skill documentation guides the installation of the markit-ai package via the bun package manager. Additionally, it describes a plugin system that allows users to install extensions directly from npm and GitHub, which involves downloading and executing third-party code.
[PROMPT_INJECTION]: The skill facilitates indirect prompt injection (Category 8) by design, as it ingests and converts content from untrusted sources into a format the agent processes.
Ingestion points: The tool is designed to read and convert content from external URLs, PDFs, Office documents, and media files as specified in the SKILL.md usage examples.
Boundary markers: The skill does not provide or mandate the use of delimiters or instructions to treat the converted output as untrusted data.
Capability inventory: The agent has the capability to execute shell commands and perform network operations to fetch external resources.
Sanitization: There is no evidence of sanitization or filtering of the output content before it is presented to the agent's context.

jb-markit