The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by ingesting and processing data from an external website (Jinri Toutiao). If the content on the target website is manipulated, it could provide malicious instructions to the agent.
Ingestion points: Untrusted data enters the agent context via the snapshot tool when capturing elements from https://www.toutiao.com as described in SKILL.md.
Boundary markers: The skill does not implement boundary markers or instructions to ignore embedded commands within the extracted suggestions.
Capability inventory: The skill uses the browser tool for navigation and the act tool for simulated user input on the target site (SKILL.md).
Sanitization: No sanitization, escaping, or validation of the extracted keyword text is performed before it is formatted for output.
[SAFE]: The skill interacts exclusively with a well-known service (www.toutiao.com) using standard browser automation tools. There is no evidence of unauthorized data exfiltration, sensitive credential access, or malicious command execution.

jinritoutiao-keyword-collector