Skip to main content
Beta: LLM Observability is now GA
aiAxonIQ
Try Playground
DocsUse cases
Get Started Free
AI Agent Runtime

Connect AI agents directly
to your telemetry plane.

Enterprise Model Context Protocol (MCP) server providing 25 forensic analysis and remediation tools for Claude Code, Cursor, and autonomous engineering agents.

Start free Try the Playground
MCP 2026-07-28 spec25 agent toolsCursor & Claude CodeCost-guard ceilingsSealed write confirmationZero tenant leakage
telemetry.aiaxoniq.net/mcp
STREAM LIVE
Model Context Protocol (MCP 2026-07-28) Agent Tool Gateway
25 Guarded Tools Active
// Invocation from Claude Code / Cursor agent:
tools/call "aiaxoniq_query_traces" { service: "checkout-api", min_latency_ms: 100 }
⚡ Telemetry Response (14ms ClickHouse rollup):
Found 3 anomalous traces. Root cause: connection timeout on downstream redis-cluster node 2. Cost guard: $0.00 / Query within rate limit.

Deep Dive

Core Capabilities & Architecture

Explore how Model Context Protocol (MCP) instruments, processes, and correlates high-cardinality telemetry at scale.

Agent Tools

Give your AI assistants real observability tools

Equip Claude Code, Cursor, and Windsurf with live production tools: get_trace, query_metrics, search_logs, investigate_incident, and get_service_health.

  • Full conformance with the Model Context Protocol 2026-07-28 specification
  • 11 read-only forensic inspection tools plus 1 human-confirmed write tool
  • Opaque cursor pagination across high-volume log streams and span trees
  • Direct integration with IDE assistants and terminal agents
Interactive Engine
Agent Tools Telemetry Engine
Real-time Stream
Capability 1
Full conformance with the Model Context Protocol 2026-07-28 specification
Capability 2
11 read-only forensic inspection tools plus 1 human-confirmed write tool
Capability 3
Opaque cursor pagination across high-volume log streams and span trees
ClickHouse 24.8 · OTLP 1.3 Native100% Ingestion Parity
Cost Guards

Prevent runaway queries from agent loops

Autonomous agent queries pass through ClickHouse cost guards that enforce strict 200,000,000 estimated-row ceilings and query deadline timeouts.

  • Pre-execution query analysis bounds ClickHouse scanning costs
  • Per-tool timeout deadlines ensure agent responsiveness
  • Redis rate limiting and token bucket quotas per service account
  • Structured query audit logging for compliance and governance
Interactive Engine
Cost Guards Telemetry Engine
Real-time Stream
Capability 1
Pre-execution query analysis bounds ClickHouse scanning costs
Capability 2
Per-tool timeout deadlines ensure agent responsiveness
Capability 3
Redis rate limiting and token bucket quotas per service account
ClickHouse 24.8 · OTLP 1.3 Native100% Ingestion Parity
Sealed Remediation

Human-in-the-loop confirmation before action

Remediation tools cannot execute unattended. The server returns a signed confirmation payload that requires an authenticated operator token before execution.

  • Sealed cryptographic request state prevents argument tampering
  • Incident mitigation actions require explicit operator sign-off
  • Audit trail logs every agent inquiry, tool selection, and outcome
  • Multi-role access control restricts agent tool execution scopes
Interactive Engine
Sealed Remediation Telemetry Engine
Real-time Stream
Capability 1
Sealed cryptographic request state prevents argument tampering
Capability 2
Incident mitigation actions require explicit operator sign-off
Capability 3
Audit trail logs every agent inquiry, tool selection, and outcome
ClickHouse 24.8 · OTLP 1.3 Native100% Ingestion Parity

Built for every team that cares about reliability

One platform, tailored to how your team actually works.

AI & DevOps engineers

Debug from your IDE

Ask Cursor or Claude Code to inspect the production trace causing a local test failure.

On-call responders

Automated triage partner

Let your autonomous agent summarize incident evidence while you join the bridge.

Platform security

Safe AI integration

Ensure AI assistants can query telemetry without data leakage or runaway database costs.

Under the hood

Transport over existing cost-guarded query APIs

The aiAxonIQ MCP server is a secure transport layer over verified internal tool registries. It introduces no raw SQL execution, extracts tenant identities from cryptographic tokens, and redacts internal identifiers before returning payloads.

MCP 2026-07-28 Go SDK200M row cost ceilingSealed confirmation token

Also in the platform

AI Insights LLM Insights Alerting

Start with MCP agent tools in 5 minutes

Free forever for up to 3 services. No credit card required.

Get started free Read the docs