The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external data that could potentially contain malicious instructions intended to influence the agent's behavior.
Ingestion points: The skill retrieves untrusted data through Amplitude:get_feedback_insights, Amplitude:get_feedback_mentions, and Searching for content (which targets annotations and documentation).
Boundary markers: The instructions lack explicit delimiters or warnings to treat ingested feedback and annotations as untrusted data, which could allow instructions hidden in feedback to be processed as directives.
Capability inventory: The skill is restricted to analytical tools within the Amplitude platform, such as retrieving chart data, deployments, and experiments. It does not have access to file-system writes, shell execution, or non-whitelisted network operations.
Sanitization: The instructions do not specify any validation, filtering, or escaping of the retrieved feedback content before it is used to synthesize hypotheses.

analyze-chart