01 / INVENTORY
MCP inventory
Bring MCP services and their tools together in one place. See which agents use them and what changes.
LLM and MCP interaction control
AI agents, LLM and MCP. Under control.
See how agents access LLMs, MCP, API and A2A services. Discover services and tools, investigate anomalies and manage access through your existing gateways.
MVP · Lab-tested · Ready to try
Collector in your environmentFree AltprobeKeep your gateway
Services, tools and connections.
sample-stack / activitySAMPLE DATA
From call to decisionGATEWAY EVENTS
MCP service discoveredINVENTORY
research-agent→tools/call→internal-search
Source: your gatewayNext: explore tools
Trigger matchedTRACE CAPTURED
agent call→tool response→AI explanation
Events in one traceNext: investigate with AI
Access managementACCESS RULES
agent→access rule→allow / block
On selected routesDecisions from the console
Workflow illustration. Sample data.LLM / MCP / API / A2A
Connect the gateway
you already use.
APISIXenvoyOpenRestyKONGtraefikHigress
What Alertflex gives you
Less guesswork.
More clarity.
Go from an inventory of MCP services and tools to understanding connections, anomalies and risks.
02 / INVESTIGATION
Anomalies and incidents
An incident flags an anomaly in an agent's calls to LLM, MCP, API or A2A services. Traces and AI Assistant help you understand why it happened.
03 / ACCESS
Access management
Manage agents' access to services from one console. Your gateways apply those decisions.
Events in one view.
OPENSEARCH SCREENS · TEST LAB DATA
From data to action
Altprobe collects.
Alertflex helps you understand.
Search events in OpenSearch. Use Alertflex to investigate traces and incidents, get AI explanations and manage access.
Your infrastructure
AI agentsYour gatewaysLLM, MCP, API and A2ASoftware inventory
AltprobeEvent collection and correlation
OpenSearchSearch and reports
AlertflexTraces and AI Assistant
Access management
Access management
↖ Access decisions are applied at the gateway
01
Connect your gateway
Altprobe collects data on MCP services, tools and agent calls through the gateway you already operate.
02
Capture a sequence of events
Set a trigger for an event that matters. When it matches, a trace is captured for investigation.
03
Get an explanation
AI Assistant helps you investigate an incident or trace and explains what happened in plain language.
04
Manage access
Restrict access when needed. View decisions and their history in the console.
Monitoring and access management can be enabled separately. Access changes take effect when settings refresh.
Different gateways. One console.
Choose your gateway and start with a ready-to-run lab. Add control to the infrastructure you already use.
APISIX
CONNECTOR ↗
Envoy
CONNECTOR ↗
OpenResty
CONNECTOR ↗
Kong OSS
CONNECTOR ↗
Traefik
CONNECTOR ↗
Higress
CONNECTOR ↗Prebuilt rules and free tools
Add context and additional checks with free open-source tools.
01 / ANOMALIES
Anomalies in gateway traffic
Inspect traffic to LLM, MCP and API services through your gateway with prebuilt Altprobe rules and OWASP CRS. Add Qwen3Guard for additional checks.
- Altprobe rules
- OWASP CRS
- Qwen3Guard
02 / CONNECTIONS
Unwanted agent requests that bypass your proxy
Detect direct connections to LLM services by correlating Suricata network events with Falco events from processes and containers.
- Suricata
- Falco
03 / SBOM
Agent software composition and vulnerabilities
Analyze agent software composition using SBOMs and find known vulnerabilities in its components. Use Trivy or combine Syft with Grype.
- Trivy
- Syft + Grype
Try it for your use case
Start with a ready-to-run lab.
See how it works with test data. Then connect your own services.
COLLECTORFREE
Start monitoring with Altprobe.
Security events, OTEL traces and agent software vulnerability findings in OpenSearch. Use Altprobe with OpenSearch for monitoring - Alertflex is not required.
- All collected OTEL traces in OTLP/JSON: built-in collection or external sources
- Agent software composition analysis using SBOMs, with vulnerability findings exported in OCSF
- Security events in OCSF for OpenSearch Security Analytics with configured field mappings and detectors
- Traffic analysis for LLM, MCP and API calls: Altprobe and OWASP CRS rules, plus correlation of Suricata and Falco IDS events
- Six ready-to-run open-source gateway labs with tests and sample traffic, plus over 70 open-source IDS rules
START WITH APISIX
git clone https://github.com/olegzhr/altprobe.git
PLATFORMGUIDED PILOT
Investigate and manage with Alertflex.
MCP service and tool inventory, agent software composition and vulnerabilities, incidents, traces and access management - in one Alertflex console.
- Traces captured by your triggers
- AI Assistant for event investigation
- Access management and decision history
- MCP services, tools and agent connections
Together, we'll find a deployment approach that fits your services and goals.
Frequently asked questions
What to know.
A quick guide to capabilities and getting started.
More questions?What can I use for free?
Altprobe, integrations and test labs are free. Alertflex deployment and support are discussed separately.
What can I see?
See LLM API calls, MCP services and tools, and interactions with API and A2A services. The gateway identifies agents by an access key or mTLS.
What is an incident?
An incident is a detected anomaly in an agent's calls to LLM, MCP, API or A2A services - such as a suspicious tool call or an unusual sequence of requests. It highlights activity that needs review.
How do traces work?
Set a trigger - a condition for capturing related events. Review the resulting trace yourself or investigate it with AI Assistant.
How do I manage access?
Allow or block access in the console. Gateways apply those decisions. Event analysis does not undo calls that have already completed.
Where is my data stored?
Deploy Alertflex in your own environment and keep control of your data. You choose whether to use AI Assistant and which model to connect - local or external. Before AI analysis, the system masks detected personal data.
How do I get started?
Run a ready-to-use test lab. Then pilot it with your own traffic - we'll help you choose where to start.
Let's discuss your use case
Know your services.
Control their interactions.
Tell us what you want to see and control. Together, we'll decide where to start.