The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external Markdown content provided by the user as raw material and re-reads the article file from the disk, creating a vulnerability surface for instructions embedded in data. Ingestion points: The agent reads the Markdown file provided as raw material and re-reads the article file from disk as specified in SKILL.md. Boundary markers: There are no explicit instructions or delimiters to distinguish user content from agent instructions. Capability inventory: The skill has the capability to read from and write to the local file system to manage the article creation process. Sanitization: No sanitization or filtering is performed on the ingested text before it is processed or written back to the file system.

writing-beats