A comprehensive framework and pattern set for implementing resilient failure handling and state recovery in AI agent workflows.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install error-recovery
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install error-recovery using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Error Recovery?
Error Recovery is a fundamental capability for building reliable autonomous systems that can survive the unpredictable nature of LLM APIs and network environments. By utilizing these Openclaw Skills, developers can implement sophisticated patterns like exponential backoff and circuit breakers to prevent cascading failures in their agentic workflows.
This skill ensures that agents can gracefully degrade their performance instead of crashing, maintaining session integrity through context compaction and automatic resets when state corruption is detected. It is essential for any production-grade implementation of AI agents where uptime and consistency are paramount.
Error Recovery Use Cases
- Managing 429 Rate Limit errors with automated exponential backoff and retry logic.
- Handling context window overflows through intelligent token compaction strategies.
- Implementing fallback model logic to ensure continuity when primary LLMs are overloaded.
- Recovering from session corruption or role ordering errors without losing task progress.
- Ensuring non-fatal tool failures do not interrupt the primary autonomous execution loop.
How Error Recovery Works
- The agent monitors execution for specific error signatures, including HTTP statuses like 429 or 503 and internal role conflicts.
- Built-in Openclaw Skills logic triggers immediate recovery actions like retries with backoff for transient network issues.
- For structural issues like context overflow, the system initiates a compaction cycle to reduce the token footprint.
- If a primary service remains unreachable, the circuit breaker pattern prevents further requests to protect system resources.
- In cases of severe state corruption, the session is reset while critical context is re-hydrated from the local MEMORY.md file.
Error Recovery Setup
To enable advanced error recovery within your project, incorporate the standardized recovery patterns into your agent's configuration. Ensure your environment is ready by verifying your Openclaw Skills setup:
# List and verify your current skill configuration
openclaw list
# The recovery logic is typically integrated into the agent runner
# Ensure your MEMORY.md file exists for state persistence
touch MEMORY.md
Error Recovery Data Schema & Taxonomy
| Data Point | Description | Recovery Action |
|---|---|---|
| 429 Rate Limit | API request frequency exceeded | Wait and retry with backoff |
| 402 Billing | Credits or quota exhausted | Switch to fallback model |
| Context Overflow | Token limit reached | Trigger context compaction |
| Role Ordering | Session state corruption | Reset session and reload memory |
| Tool Failure | Error in external tool execution | Log error and continue execution |
Persistent metadata is stored in a structured taxonomy within the agent session to track retry counts and circuit breaker states.
Error Recovery Advanced Features
- Multi-tier model fallback chains allowing seamless transitions from expensive to cost-effective LLMs.
- Customizable exponential backoff strategies to handle high-traffic API environments.
- Automated session reset hooks that preserve persistent memory checkpoints even after a fatal crash.
- Non-fatal tool execution wrappers that prevent individual plugin failures from stalling the entire agent workflow.
- Integrated circuit breaker state management to protect upstream API usage and costs.
SKILL.md
Loading
Related Openclaw Skills
OpenClaw Hook Examples
A comprehensive developer guide for implementing and utilizing lifecycle hooks to intercept, audit, and modify AI agent behavior within the OpenClaw ecosystem.
Tool Enhancement
A comprehensive suite of over 55 professional-grade tools designed to extend the capabilities of AI agents within the Openclaw Skills ecosystem.
AWS Knowledge
An expert cloud assistant providing real-time access to AWS documentation, architecture patterns, and standard operating procedures.
Text to Video Maker Free
A sophisticated AI skill that converts written scripts, articles, and ideas into structured video content with scenes, captions, and visual flow.
Audit Log Hook
A robust logging mechanism that records every tool interaction for security and performance analysis.
Compact Summary
A structured memory compression system that optimizes context windows by summarizing conversation history into a machine-readable format.
Featured*
