The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill utilizes the images.upload tool to transmit local image files to the Ghost server and uses ghost_docs to fetch information from external documentation. These are primary functions of the skill and target the configured Ghost infrastructure.
[PROMPT_INJECTION]: The workflow involves reading existing blog content via posts.read and searching documentation via ghost_docs. These are indirect prompt injection surfaces where untrusted data from external sources could contain malicious instructions designed to influence the agent's behavior. No explicit boundary markers or sanitization steps are defined for these tool outputs.
[COMMAND_EXECUTION]: Instructions guide the agent to write JSON payloads to absolute file paths (using blockFile) to manage long posts. While this is an intended feature for handling large datasets, it relies on the agent's execution environment to enforce filesystem boundaries.
[PROMPT_INJECTION]: The skill allows for the use of html and embed blocks. While described as a last resort, these blocks enable the insertion of raw HTML into the Ghost platform, which is a standard feature but represents a vector for cross-site scripting (XSS) if the agent is directed to include malicious scripts.

ghost-writing