The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill documentation encourages downloading installation resources from a remote GitHub repository (raw.githubusercontent.com/inference-sh/skills/refs/heads/main/cli-install.md) to set up the execution environment.
[EXTERNAL_DOWNLOADS]: The skill instructions include the installation of a non-standard Node.js package (belt-sh/cli) via npx to enable its core functionality.
[COMMAND_EXECUTION]: The skill utilizes the belt CLI to run remote applications for posting content to social media and generating images, which involves executing shell commands to interact with external services.
[INDIRECT_PROMPT_INJECTION]: The skill uses tools that ingest untrusted data from the internet, such as search results and web page content, which could be used to influence the agent's behavior.
Ingestion points: Data fetched via tavily/search-assistant and infsh/agent-browser (SKILL.md).
Boundary markers: No specific delimiters or safety warnings are provided for the agent when processing this external content.
Capability inventory: The agent has the ability to post content to Twitter/X via x/post-create based on the ingested information.
Sanitization: The skill does not explicitly show sanitization or validation of the text retrieved from external sources before it is used to generate social media posts.

twitter-thread-creation