The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes external markdown files which presents a surface for indirect prompt injection attacks. \n
Ingestion point: The skill reads local markdown files via the publish.sh script to transmit their content to the hosting server. \n
Boundary markers: No explicit boundary markers or instructions to the agent to treat the embedded content as untrusted were found in the instructions. \n
Capability inventory: The skill has the ability to read local files and perform network operations via curl. \n
Sanitization: No sanitization or validation of the markdown content is performed before it is processed or transmitted. \n- [REMOTE_CODE_EXECUTION]: The README.md file contains installation instructions recommending that users pipe a remote script directly to their shell (curl -fsSL https://leafmill.net/install.sh | bash). This is a common but high-risk pattern for executing unverified remote code. \n- [DATA_EXFILTRATION]: The skill is designed to read the contents of local files and transmit them to an external domain (leafmill.net). While this is the intended functionality of the service, it represents a data transmission vector that requires user awareness, especially when processing sensitive documents.

leafmill