Can I distinguish between rate limits and model errors?

Yes, the skill automatically classifies errors into categories like transient (rate limits, timeouts) and permanent (invalid requests, context length exceeded).

What is fallback tracking?

It monitors when a system switches from a primary model to a backup model due to failures, tracking the trigger reason, the model chain, and the impact on output quality.

Does it support popular frameworks like LangChain?

Yes, it provides implementation patterns for integrating with LangChain callbacks and Tenacity to ensure robust retry and error handling.

How does this skill help reduce LLM costs?

It identifies permanent errors early to prevent expensive, useless retries and tracks 'wasted' tokens from failed requests, allowing you to optimize your resource allocation.

Error and Retry Tracking

Name: Error and Retry Tracking
Author: nexus-labs-automation

bynexus-labs-automation

0•

分析と監視

Instruments and monitors AI agent error handling, retry logic, and fallback strategies to improve system reliability and observability.

This skill provides a comprehensive framework for tracking the lifecycle of failures within AI agent workflows. It enables developers to distinguish between transient and permanent errors, monitor retry success rates, visualize fallback paths, and manage rate limits effectively. By providing deep insights into failure patterns and recovery behaviors, it helps optimize LLM performance, reduce wasted token costs, and implement resilient circuit breaker patterns for more stable production deployments.

主な機能

01Granular tracking of retry attempts, strategies, and delays

02Circuit breaker pattern implementation for cascading failures

03Classification of transient vs. permanent LLM errors

04Rate limit monitoring and preemptive wait handling

05Observability for model fallback chains and quality impact

060 GitHub stars

ユースケース

01Optimizing retry strategies to balance response latency and token costs

02Debugging agentic workflows that stall or fail in production environments

03Monitoring provider health and rate limit exhaustion across multiple LLM models

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add nexus-labs-automation/agent-observability error-retry-tracking

For use in Claude.ai and ChatGPT

Download Skill