Does it support saving partial results?

Yes, it tracks success and failure across multiple agents and can proceed with a consensus analysis as long as a minimum threshold of successful responses is met.

How does this skill handle model timeouts?

It monitors execution time and provides automated options to retry with increased limits, skip the stalled model, or cancel the workflow gracefully to avoid hanging processes.

What happens if all external models fail?

The skill triggers a fallback mechanism that switches to embedded models or local alternatives, ensuring the user still receives a result even if external services are unavailable.

Can it recover from API rate limits (429 errors)?

Yes, it identifies 429 status codes and implements waiting periods based on 'Retry-After' headers or default exponential backoff strategies before attempting a retry.

How does it handle user cancellation via Ctrl+C?

It implements a cleanup strategy that stops all running sub-processes gracefully and saves any partial progress to disk before exiting.

Error Recovery Workflow

Name: Error Recovery Workflow
Author: tianzecn

bytianzecn

•

Productivity & Workflow

Implements resilient error-handling patterns and fallback strategies for multi-agent AI workflows to ensure reliable execution.

The Error Recovery skill provides a robust framework for managing the inherent unpredictability of multi-agent systems and external AI dependencies. It equips Claude with battle-tested strategies for handling timeouts, API failures, and rate limits through intelligent retry logic, exponential backoff, and graceful degradation. By enabling partial success processing and structured user cancellation handling, this skill ensures that complex workflows remain stable and productive even when individual components encounter issues, making your AI integrations production-ready.

Key Features

01Graceful degradation patterns for multi-model consensus and partial success

02Intelligent retry strategies with exponential backoff for transient API failures

03Automated timeout management for long-running external model responses

04Structured handling for rate limits (429) and authentication errors (401)

051 GitHub stars

06Robust cleanup and partial result saving upon user cancellation (SIGINT)

Use Cases

01Ensuring workflow continuity by falling back to local or embedded models during network instability

02Rescuing multi-model code reviews when specific providers time out or fail

03Managing API rate limits and quotas during large-scale automated refactoring

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add tianzecn/myclaudecode error-recovery

For use in Claude.ai and ChatGPT

Download Skill