The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill's installation documentation includes a command to download and execute a shell script directly from a remote GitHub repository.
Evidence: The command curl -fsSL https://raw.githubusercontent.com/xdevplatform/xurl/main/install.sh | bash is recommended for Linux and macOS installation in SKILL.md.
[EXTERNAL_DOWNLOADS]: The skill suggests using external package managers to install the required CLI tool, fetching code from remote registries.
Evidence: Installation instructions in SKILL.md include npm install -g @xdevplatform/xurl and go install github.com/xdevplatform/xurl@latest.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from the X platform, which could contain malicious instructions designed to influence the agent's behavior.
Ingestion points: Untrusted content enters the agent's context through search results, timeline retrieval, mentions, and direct messages, as described in SKILL.md.
Boundary markers: The skill lacks technical delimiters for external content in its execution logic, though SKILL.md provides instructional safety guidelines for the agent to confirm user intent before performing write operations.
Capability inventory: The skill possesses significant 'write' capabilities, including posting, replying, sending direct messages, and following or blocking accounts, which could be targeted by an injection attack.
Sanitization: While the tool returns structured JSON output, the underlying message content is processed by the LLM without specific sanitization against embedded instructions.

xurl