How does it handle model timeouts?

It monitors execution time against a threshold and provides options to retry with an increased timeout, skip the specific model, or cancel the workflow entirely.

Can it handle partial failures in multi-model tasks?

Yes, it implements logic to determine if enough models succeeded to reach a consensus, proceeding with available data while notifying the user of specific failures.

What happens if an API key is missing or invalid?

The skill detects 401 Unauthorized errors, logs the specific cause, and automatically falls back to embedded Claude models to ensure the task can still complete.

What triggers the Error Recovery skill?

The skill is triggered by specific keywords such as 'error', 'timeout', 'retry', 'failure', and 'fallback' to manage disruptions in multi-agent workflows.

Does it support exponential backoff?

Yes, for network-related errors, the skill uses an exponential backoff strategy (e.g., 1s, 2s, 4s) to avoid overwhelming services while attempting recovery.

Resilient Workflow Error Recovery

Name: Resilient Workflow Error Recovery
Author: involvex

byinvolvex

Productivity & Workflow

Implements robust patterns for handling timeouts, API failures, and partial successes within complex multi-agent AI workflows.

About

The Error Recovery skill provides a comprehensive framework for building resilient Claude Code workflows that gracefully handle the inevitable failures of external dependencies. It equips agents with sophisticated strategies for timeout management, exponential backoff retries for transient API errors, and graceful degradation when partial successes occur. By implementing these battle-tested patterns, developers ensure their multi-agent systems remain production-ready, providing users with consistent results and clear communication even when individual models or services encounter issues like rate limits or network latency.

Key Features

Structured handling of user cancellations to ensure clean process termination
0 GitHub stars
Intelligent fallback to embedded models when external providers are unavailable
Graceful degradation patterns to maintain workflow continuity during partial successes
Exponential backoff retry strategies for transient network and 500-level API errors
Automated timeout detection and recovery for slow external AI models

Use Cases

Building production-grade automations that must survive 429 rate limits and 500 server errors
Managing multi-model code reviews where specific providers may experience latency or outages
Protecting long-running background tasks from data loss during unexpected user interruptions

About

Key Features

Structured handling of user cancellations to ensure clean process termination
0 GitHub stars
Intelligent fallback to embedded models when external providers are unavailable
Graceful degradation patterns to maintain workflow continuity during partial successes
Exponential backoff retry strategies for transient network and 500-level API errors
Automated timeout detection and recovery for slow external AI models

Use Cases

Building production-grade automations that must survive 429 rate limits and 500 server errors
Managing multi-model code reviews where specific providers may experience latency or outages
Protecting long-running background tasks from data loss during unexpected user interruptions