How does the cost tracking work?

It includes a configurable pricing map for Mistral models like Small, Large, and Embed, calculating estimated costs based on actual token usage in real-time.

What kind of alerts are included?

It includes alert rules for high error rates (above 5%), P95 latency spikes, rate-limiting detection, high hourly costs, and service unavailability.

What specific metrics does this skill track for Mistral?

The skill tracks total API requests, latency histograms, input/output token counts, error types, and estimated USD costs per model.

Does this require a specific observability backend?

Yes, it is optimized for use with Prometheus for metrics collection, OpenTelemetry for tracing, and Grafana for visualization.

Can I use this with existing TypeScript projects?

Absolutely. The skill provides fully typed instrumented client wrappers and middleware examples designed specifically for TypeScript and Node.js environments.

Mistral AI Observability

Name: Mistral AI Observability
Author: jeremylongshore

byjeremylongshore

•

1,538

•

Analytics & Monitoring

Implements professional-grade monitoring, distributed tracing, and automated alerting for Mistral AI integrations.

This skill provides a comprehensive framework for monitoring Mistral AI operations within production environments. It automates the setup of Prometheus metrics for tracking request latency, token usage, and API costs, while integrating OpenTelemetry for deep distributed tracing. By providing structured logging patterns, ready-to-use Grafana dashboard configurations, and AlertManager rules, it ensures developers can maintain high reliability and strict cost-control over their AI-powered applications.

Key Features

011,538 GitHub stars

02Structured logging templates for standardized auditing of AI operations

03Pre-configured AlertManager rules for high error rates, latency, and rate-limiting

04Automated Prometheus metrics for token usage, latency, and estimated costs

05Grafana dashboard JSON templates for immediate visual monitoring and reporting

06OpenTelemetry integration for end-to-end distributed tracing across AI services

Use Cases

01Setting up proactive alerts for API downtime or unexpected model behavior

02Debugging performance bottlenecks in AI workflows using distributed traces

03Monitoring production LLM costs and token consumption across different Mistral models

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add jeremylongshore/claude-code-plugins-plus-skills mistral-observability

For use in Claude.ai and ChatGPT

Download Skill