The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill recommends an installation method that downloads a shell script from a remote URL and pipes it directly into bash for execution.
Evidence: curl -fsSL https://raw.githubusercontent.com/xdevplatform/xurl/main/install.sh | bash in SKILL.md.
This practice facilitates arbitrary code execution from a remote source on the user's system.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted, user-generated content from the X (Twitter) API, creating a vulnerability to indirect prompt injection.
Ingestion points: The agent reads external data through commands such as xurl search, xurl timeline, xurl mentions, and xurl dms in SKILL.md.
Boundary markers: The skill does not provide clear delimiters or instructions to treat the text content of tweets or messages as potentially untrusted data.
Capability inventory: The skill enables the agent to take significant actions based on potentially malicious input, including posting tweets (xurl post), sending direct messages (xurl dm), and replying to existing posts (xurl reply).
Sanitization: No sanitization or validation protocols are defined for the incoming API data before it is processed by the agent.
[COMMAND_EXECUTION]: The skill relies on the execution of shell commands through the xurl CLI utility to interact with the API.
Evidence: Multiple command examples and a 'Raw API Access' section in SKILL.md that allow for arbitrary request construction.

xurl