What happens if a specific AI model times out?

The skill monitors execution time and provides options to retry with a longer limit, skip the failing model while continuing with others, or cancel the workflow gracefully.

How does this skill handle API rate limits?

It detects 429 status codes and automatically waits according to the 'Retry-After' header or a 60-second default before attempting a retry, ensuring compliance with provider limits.

Does it support graceful cancellation?

Yes, the skill captures interruption signals like Ctrl+C to stop sub-agents properly and save any partial progress or logs before the process exits.

Can the workflow continue if only some models succeed?

Yes, it implements partial success strategies that adapt consensus logic to the number of successful responses, allowing the process to move forward if a minimum threshold is met.

Error Recovery & Resilience

Name: Error Recovery & Resilience
Author: MadAppGang

byMadAppGang

•

230

•

Developer Tools

Implements robust error handling, intelligent retries, and graceful degradation patterns for multi-agent AI workflows.

The Error Recovery skill provides a standardized framework for making multi-agent AI workflows production-ready by handling the inevitable failures of external dependencies. It implements battle-tested patterns for managing model timeouts, API authentication errors, rate limiting, and network instability. By utilizing partial success strategies and flexible consensus logic, this skill ensures that your automation pipelines remain resilient, allowing workflows to complete with available data rather than failing entirely when a single component encounters an issue.

Key Features

01Automated timeout detection and recovery for external model responses

02Intelligent retry strategies with exponential backoff for transient network errors

03230 GitHub stars

04SIGINT handling for clean workflow termination and partial result preservation

05Comprehensive API error handling for 401, 429, and 500 status codes

06Graceful degradation logic for maintaining workflow continuity during partial failures

Use Cases

01Automating long-running background tasks that require resilience against intermittent API rate limits

02Implementing fallback mechanisms to local or embedded models when external API quotas are exhausted

03Maintaining multi-model consensus workflows even when specific providers experience downtime

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add madappgang/claude-code error-recovery

For use in Claude.ai and ChatGPT

Download Skill