Fetch a single AI insights run by ID, scoped to the authenticated team.

The dashboard_url field is a direct link to the Insights page in the Revenium UI for this run.

Full mode (default, slim=false): Returns all fields including findingsJson, recommendationsJson, and batchMetadataJson (raw JSON strings requiring JSON.parse), plus pre-parsed equivalents parsedFindings, parsedRecommendations, and parsedBatchMetadata for MCP consumers. Use findingsCount and recommendationsCount for quick metadata checks without parsing.

Slim mode (slim=true): Omits findingsJson, recommendationsJson, parsedFindings, and parsedRecommendations. Instead returns recommendationsSummary: [{ id, title, rank, severity, monthlyImpact, confidenceBand, category }] — a compact ranked list safe to inline in MCP tool responses without exceeding context budgets. batchMetadataJson and parsedBatchMetadata are still included. Use findingsTruncated and findingsConsideredCount to detect whether findings were capped before synthesis.