The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the conversation or external files and interpolates it directly into markdown documents.
Ingestion points: Content is gathered from conversation history or user-specified files as described in Step 2 of SKILL.md.
Capability inventory: The skill possesses file-write capabilities and the ability to execute pandoc via Bash.
Sanitization: The instructions lack specific guidance on sanitizing or escaping content before it is processed by pandoc or used in YAML frontmatter.
Boundary markers: No specific delimiters or warnings are used to prevent untrusted content from overriding document structure or metadata.
[COMMAND_EXECUTION]: The skill uses the Bash tool to execute pandoc and file management commands with arguments derived from untrusted content.
Dynamic Argument Construction: Output filenames and metadata fields (such as document titles) are generated based on the content of the document. If the agent does not strictly validate these strings, they could be used to manipulate command line arguments passed to pandoc, mkdir, or open.

generate-document