The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references external resources for tool installation, including instructions to add the belt-sh/cli skill and a link to a remote installation guide on GitHub (inference-sh/skills).
[COMMAND_EXECUTION]: The skill provides numerous examples of executing the belt CLI tool to perform tasks like social media posting, image generation, and web browsing.
[DATA_EXFILTRATION]: The skill includes functionality to send text and media to the X (Twitter) platform and to communicate with external research APIs like Tavily.
[PROMPT_INJECTION]: The skill contains an attack surface for indirect prompt injection via tools that fetch external data.
Ingestion points: External data is ingested through the tavily/search-assistant and infsh/agent-browser tools as described in SKILL.md.
Boundary markers: The prompt templates do not include markers or instructions to treat external data as untrusted.
Capability inventory: The skill allows the agent to post content to an external service (x/post-create in SKILL.md).
Sanitization: The instructions lack steps for sanitizing or verifying the integrity of retrieved information before it is repurposed into social media posts.

twitter-thread-creation