WaveGuard
Scan time series or metrics through a physics-simulation anomaly API so your agent flags outliers in production or analytics pipelines.
Overview
WaveGuard is an Operate-phase MCP server that exposes a physics-simulation anomaly detection API to scan data for outliers over a hosted remote endpoint.
What is this MCP server?
- Anomaly detection API powered by physics simulation to scan data for outliers
- Hosted remote MCP at Modal-run FastAPI v2 endpoint (WaveGuard 3.1.0)
- GitHub repository LFMAnomalyDetection and WaveGuardClient documentation links
- Streamable-http MCP transport for agent-driven scans
- Focused outlier detection rather than full observability suite
- Server version 3.1.0 (WaveGuard)
- Remote MCP hosted on Modal FastAPI app (registry v2/mcp path)
- GitHub: https://github.com/gpartin/LFMAnomalyDetection
What problem does it solve?
Production and analytics data can drift or spike without you having a lightweight, agent-callable way to detect outliers beyond static thresholds.
Who is it for?
Technical solo builders who already emit metrics or logs as data series and want experimental, simulation-based anomaly checks via MCP.
Skip if: Non-technical founders needing turnkey dashboards, teams wanting full APM with traces, or use cases with no quantified time-series data.
What do I get? / Deliverables
After registering the WaveGuard MCP remote, your agent can run simulation-backed anomaly scans and surface outlier candidates for your review or alerting hooks.
- Agent-triggered outlier scans via WaveGuard anomaly detection API
- Hosted MCP v2 integration path documented in registry and GitHub client repo
- Operational hook for flagging anomalies before they become user-facing incidents
Recommended MCP Servers
Journey fit
Outlier detection is an ongoing production discipline once data flows live and you need to catch drift, fraud spikes, or sensor faults early. Monitoring subphase matches APIs whose purpose is spotting anomalies rather than building features or launching campaigns.
How it compares
Physics-simulation outlier API over MCP, not a hosted log aggregator or a unit-test skill.
Common Questions / FAQ
Who is com.emergentphysicslab/waveguard for?
It is for developers and indie operators who use AI agents and want to call an anomaly detection API when reviewing metrics, experiments, or operational data.
When should I use com.emergentphysicslab/waveguard?
Use it in Operate monitoring when you need to scan datasets for outliers, investigate suspicious spikes, or prototype alert logic before a full observability buy.
How do I add com.emergentphysicslab/waveguard to my agent?
Add the published streamable-http remote MCP URL from the registry (WaveGuard FastAPI v2 MCP on Modal) in your MCP client and follow WaveGuardClient docs on GitHub for payload expectations.