The Agent Skills Directory

[SAFE]: The skill is authored by Amplitude and uses internal platform tools to fetch session data and interaction timelines. It does not attempt to access sensitive system files, hardcode credentials, or execute arbitrary commands.
[INDIRECT_PROMPT_INJECTION]: The skill ingests interaction timelines (clicks, navigations, inputs) which originate from external users. While this constitutes an untrusted data surface, the skill lacks high-privilege tools (such as shell access, file writing, or general network requests) that would be required to exploit an injection. The impact of any potential injection is limited to affecting the generated reproduction steps, which are intended for human review.

debug-replay