The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's primary function involves ingesting and processing arbitrary plaintext files, which introduces a potential surface for indirect prompt injection if the files contain malicious instructions.
Ingestion points: The workflow described in SKILL.md and the parameters in parameters.md instruct the agent to read user-provided text files (#file) for conversion.
Boundary markers: The instructions lack explicit delimiters or strict boundary markers to isolate the ingested file content from the agent's execution context. However, considerations.md provides a partial mitigation by instructing the agent to ignore process-related keywords like exit, kill, or sleep found within the text.
Capability inventory: The agent has the ability to read local files and perform file system operations such as copying (cp or copy) to manage the conversion process, as detailed in examples.md and parameters.md.
Sanitization: There are no specific instructions for the agent to sanitize or escape the content of the documentation before processing it, relying on the agent to distinguish content from instructions during the formatting task.

convert-plaintext-to-md