The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the Typefully API, such as draft text and comments, creating a potential surface for indirect prompt injection where malicious instructions could be embedded in social media content.
Ingestion points: The scripts/typefully.js tool fetches external content via drafts:get and comments:list commands.
Boundary markers: The skill documentation instructs the agent to preserve specific <typ:comment-thread> anchors for comments, which provides structural boundaries, though the main post content lacks explicit safety delimiters.
Capability inventory: The skill allows the agent to publish content to multiple social media platforms, upload media, and modify its own local configuration files.
Sanitization: The script includes filename sanitization for media uploads but does not sanitize the text content of drafts or comments before they are processed by the agent.
[DATA_EXPOSURE]: The skill manages a TYPEFULLY_API_KEY by reading from and writing to specific local configuration files (~/.config/typefully/config.json and .typefully/config.json).
These operations represent standard CLI secret management practices.
The instructions include explicit guardrails forbidding the agent from searching for credentials in sensitive locations like .env files, Keychain, or system logs, which reduces the risk of credential exposure.

typefully