canonical: https://jentic.com/apis/amazonaws.com/lookoutmetrics # AWS Amazon Lookout for Metrics Jentic publishes the only available OpenAPI specification for Amazon Lookout for Metrics, keeping it validated and agent-ready. Amazon Lookout for Metrics uses machine learning to detect anomalies in business and operational metrics - sales drops, conversion changes, traffic spikes - without requiring data science expertise. Anomaly detectors ingest data from sources such as Amazon S3, Amazon Redshift, Amazon CloudWatch, and Amazon RDS, then surface deviations along with severity scores and grouped contributing dimensions. Alerts route findings to channels like Amazon SNS or AWS Lambda so downstream systems can respond automatically. ## For AI agents Detect anomalies in metric time series, manage anomaly detectors, and route findings to alert channels for automated response. ## Scope Does not handle log analytics, root cause analysis across distributed traces, or general-purpose forecasting - use for time-series metric anomaly detection only. ## Capabilities - Create anomaly detectors that monitor metric sets sourced from S3, Redshift, CloudWatch, or RDS - Activate, deactivate, or back-test anomaly detectors against historical metric data - Configure metric sets with measure and dimension fields for grouped anomaly analysis - Surface anomaly groups ranked by severity and inspect contributing dimensions - Provide human feedback on detected anomalies to refine detector accuracy - Route detected anomalies to SNS topics or Lambda functions through alert configurations - Tag detectors and metric sets for cost allocation and access control ## Use cases ### Revenue and Conversion Anomaly Detection E-commerce and SaaS teams use Lookout for Metrics to spot unexpected drops in revenue, sign-ups, or conversion rates without writing forecasting code. The service builds models per metric set, learns seasonality, and surfaces anomalies grouped by contributing dimensions such as product, region, or channel. Setup is point-and-click against an S3 or Redshift data source and produces results within hours of activation. Example prompt: Create an anomaly detector for daily revenue with a one-day frequency, attach an S3-backed metric set with dimensions product_category and region, and activate the detector ### Operational Telemetry Monitoring DevOps and SRE teams attach Lookout for Metrics to CloudWatch metrics to find anomalies in latency, error rates, and throughput across services. Detectors group anomalies by service or instance dimension, helping engineers locate root causes faster than fixed-threshold alarms. The CloudWatch source connector ingests metrics directly so no ETL is required. Example prompt: Configure a metric set sourced from CloudWatch latency metrics, attach it to an existing detector, and create an alert that triggers a Lambda function for anomalies above severity 70 ### Marketing Campaign Performance Tracking Marketing analytics teams monitor click-through rates, ad spend, and acquisition costs across campaigns. Lookout for Metrics groups anomalies by campaign and channel dimensions, flagging when a single ad creative collapses while overall metrics stay healthy. Feedback loops let analysts mark false positives so the model adapts. Example prompt: List anomaly groups for the marketing detector created in the last 7 days, then post user feedback marking the top result as anomaly type relevant ### AI Agent Anomaly Triage Workflows AI agents call Lookout for Metrics through Jentic to retrieve fresh anomaly groups, fetch contributing dimensions, and decide whether to open a ticket, page an on-call engineer, or annotate a dashboard. Jentic isolates the AWS Signature v4 credentials and exposes the detector and metric-set operations as discoverable tools so the agent can act without learning the SDK. Example prompt: Search Jentic for 'detect a metric anomaly', load the ListAnomalyGroupSummaries operation for detector arn:aws:lookoutmetrics:..., and execute it for the last 6 hours ## Key endpoints | Method | Path | Description | | --- | --- | --- | | POST | /CreateAnomalyDetector | Create a new anomaly detector | | POST | /CreateMetricSet | Attach a metric set to a detector | | POST | /ActivateAnomalyDetector | Activate a detector to begin training and ingestion | | POST | /BackTestAnomalyDetector | Run a detector against historical data | | POST | /ListAnomalyGroupSummaries | List detected anomaly groups for a detector | | POST | /CreateAlert | Create an alert that routes anomalies to SNS or Lambda | | POST | /PutFeedback | Submit human feedback on a detected anomaly | ## Key resources - **AnomalyDetectors** — Create, activate, deactivate, and back-test anomaly detectors that learn baselines from metric data - **MetricSets** — Define metric, dimension, and timestamp fields for the data the detector ingests - **AnomalyGroups** — Inspect detected anomalies, their severity scores, and contributing dimensions - **Alerts** — Route anomalies to SNS topics or Lambda functions based on severity thresholds - **Feedback** — Submit human feedback on anomalies so the detector adapts over time ## Why Jentic - **Setup:** Wiring Lookout for Metrics by hand means implementing AWS SigV4 signing, targeting the regional host such as lookoutmetrics.us-east-1.amazonaws.com, and handling account and per-detector quotas with your own retry logic. Through Jentic you install once, import Amazon Lookout for Metrics from the API Directory, store the AWS access key once, and your agent calls it. - **Permission scoping:** Lookout for Metrics passes the detector and metric-set ARN in the request body rather than the URL path, so scoping is by operation, not by path prefix. Limit the agent to the operations it needs, such as ListAnomalyGroupSummaries and DescribeAnomalyDetector for read-only triage, so mutating ones like DeleteAnomalyDetector are not included unless you add them. - **Credential handling:** Your AWS access key is stored once, encrypted, by your own Jentic One instance and used to sign each Lookout for Metrics request with SigV4 at execution time. The raw secret never enters the agent's prompt, logs, or context. - **Discovery method:** Agents search Jentic by intent such as 'detect a metric anomaly' or 'list anomaly groups', and Jentic returns the matching Lookout for Metrics operation with its input schema so the agent calls the right endpoint without browsing the AWS reference. ## Related APIs - **Amazon CloudWatch** — CloudWatch is the metric source many Lookout detectors ingest from - **Amazon Forecast Query** — Forecast produces probabilistic forecasts; Lookout detects anomalies after the fact - **Amazon SageMaker** — SageMaker offers full custom ML model training; Lookout is a managed anomaly service ## FAQ ### Why is there no official OpenAPI spec for Amazon Lookout for Metrics? AWS does not publish an OpenAPI specification. Jentic generates and maintains this spec so that AI agents and developers can call Amazon Lookout for Metrics via structured tooling. It is validated against the live API and kept up to date. Get started with Jentic One, the self-hosted execution layer. ### What authentication does the Amazon Lookout for Metrics API use? The API uses AWS HMAC request signing (Signature Version 4) with an access key ID and secret access key scoped via IAM. Through Jentic, AWS credentials are stored encrypted in the vault and never enter the agent's prompt context - the agent receives a scoped session that signs each call. ### Can I create and activate an anomaly detector with the Amazon Lookout for Metrics API? Yes. POST /CreateAnomalyDetector defines a detector with frequency and KMS settings, POST /CreateMetricSet attaches the data source and metric definitions, and POST /ActivateAnomalyDetector starts model training and ingestion. Back-testing against historical data is available via POST /BackTestAnomalyDetector. ### What are the rate limits for the Amazon Lookout for Metrics API? AWS does not document hard request-per-second limits for Lookout for Metrics control-plane endpoints in the public spec. Quotas apply to detectors per account, metric sets per detector, and ingested data volume - see the AWS service quotas console for current values for your region. ### How do I retrieve recent anomaly groups through Jentic? Search Jentic for 'list anomaly groups' to discover POST /ListAnomalyGroupSummaries, load its schema with the Jentic Python SDK (pip install jentic), and execute it with the detector ARN and the desired time range. The MCP tool name is aws_list_anomaly_groups. ### Is Amazon Lookout for Metrics free? No. Lookout for Metrics charges per metric analyzed per month with a free tier for new accounts in the first month. Detector creation and metric-set configuration are free; you pay for the metrics actively monitored. See the AWS pricing page for the current rate per metric. ### Can I limit what my agent is allowed to do with the Amazon Lookout for Metrics API? Yes. Because Jentic One is self-hosted, your own rules decide which operations and credentials the agent may use. Lookout for Metrics passes the detector and metric-set ARN in the request body rather than the URL path, so scoping is by operation rather than by path prefix. You can restrict the agent to read-only triage operations such as ListAnomalyGroupSummaries and DescribeAnomalyDetector, leaving mutating operations like DeleteAnomalyDetector out unless you explicitly add them.