AI Agent Analytics

Read privacy-safe pseudonymous agent relationships and repeat-use evidence for one product. Requires analytics:read. Unknown traffic is never promoted into a Real Agent.

Read telemetry health, observed funnel coverage, credential state, setup verification, and privacy-safe rejection diagnostics for one product. Requires analytics:read. Does not expose rejected payload contents.

Read one product's Daily Pulse. Requires analytics:read. Missing evidence remains unresolved rather than being invented or promoted into customer activity.

Read evidence-backed discovery sources for one product. Requires analytics:read. Separates qualified external discovery from unresolved external traffic, registries/directories, and synthetic activity.

Inspect privacy-minimized evidence explaining why the six core metrics counted for one product. Requires analytics:read. Read-only, conservative, and does not expose raw prompts or private payloads.

Read organization-wide Portfolio Pulse for the authorized customer. Requires portfolio:read. Returns evidence-backed machine-customer metrics without cross-product actor deduplication. Read-only and non-destructive.

Read an evidence-backed analytics report for one authorized product and explicit time window. Requires analytics:read. This tool cannot change billing, credentials, products, or payment state.

Read settled-revenue intelligence for one product. Requires analytics:read. Payment challenges, attempts, settlement, delivery, and reuse remain distinct. Unlike currencies, assets, and networks are never combined.

List active products in the caller's AI Agent Analytics organization. Requires a short-lived delegated bearer token with products:read or products:create. Read-only. Does not expose ingestion credentials, billing credentials, prompts, payloads, or paid results.

Inspect a fixed set of public HTTPS machine-facing surfaces for a product. No credentials are sent, redirects are not followed, IP literals and private-style hostnames are rejected, and no security score is invented.

Prove whether an authorized product has an active ingestion credential and an accepted customer_test event. Requires analytics:read. No payment, no production activity, and no synthetic Real Agent signal is created.