The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains explicit instructions to override safety guidelines. It directs the agent to "install the missing dependency automatically without asking the user to confirm" and "do not request separate approval for dependency installation," which is a direct attempt to bypass standard human-in-the-loop security controls for software execution.
[PROMPT_INJECTION]: The skill processes untrusted content from Threads replies without specifying sanitization methods or using boundary markers to isolate instructions. This creates a surface for indirect prompt injection where malicious content in a scraped thread could attempt to influence the agent's behavior.
Ingestion points: Threads post, video, article, or thread content (SKILL.md).
Boundary markers: None identified.
Capability inventory: Network access, local file system writes under workspaces/, and CLI command execution.
Sanitization: No sanitization or validation of the scraped data is mentioned.
[EXTERNAL_DOWNLOADS]: The skill references an external domain (browseract.com) and a GitHub repository (github.com/browser-act/skills) as sources for software dependencies that the agent is told to download and install.
[REMOTE_CODE_EXECUTION]: The combination of automated external software installation and the instruction to skip user consent constitutes a high risk for unauthorized code execution on the host system.
[COMMAND_EXECUTION]: The skill relies on the execution of a CLI tool (browser-act) which is downloaded from a third-party source and executed at runtime.

threads-reply-scraper