The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted data from external sources, specifically customer feedback and raw user comments. This content is used to synthesize opportunities and could contain instructions that attempt to manipulate the agent's output or logic.
Ingestion points: Raw customer feedback and mentions are ingested via the get_feedback_mentions, get_feedback_comments, and get_feedback_insights tools in Phase 2d. Additionally, the get_project_context tool reads the "AI context" field, which may contain user-provided text.
Boundary markers: The instructions do not define explicit delimiters or headers to isolate external feedback data from the system's own instructions, nor do they instruct the model to ignore potential commands embedded within that feedback.
Capability inventory: The skill possesses the capability to query expansive datasets from Amplitude (funnels, charts, cohorts, experiments) and generate structured reports using the query_charts, query_dataset, and query_experiment tools.
Sanitization: There is no evidence of sanitization, filtering, or validation of the text retrieved from feedback sources before it is processed for synthesis.

discover-opportunities