failure-analyzer

What This Skill Does

The Failure Analyzer is a robust meta-skill designed for OpenClaw AI agents to facilitate continuous improvement and rigorous root-cause analysis. It operates as an observational layer that tracks failed tasks, analyzes them using the '5 Whys' framework, and stores the resulting lessons in a structured knowledge base. By identifying systemic issues—whether they are technical bugs, operational inefficiencies, or strategic missteps—the agent effectively creates an institutional memory, ensuring that once a mistake is recognized, the probability of repetition is significantly reduced through automated pattern recognition and preventive warning triggers.

Installation

To integrate this skill into your OpenClaw environment, execute the following command in your terminal or command-line interface:

clawhub install openclaw/skills/skills/sa9saq/failure-analyzer

Ensure that your environment has sufficient write permissions to the /moltworker/data/learning/ directory, as the skill requires this space to maintain its internal JSON-based learning database.

Use Cases

• Automated Bug Resolution: When an API integration fails, the system automatically runs a post-mortem analysis to determine if the issue was a transient network flicker or a fundamental logic error.
• Operational Efficiency: If a recurring reporting task consistently misses its KPI, the skill identifies the procedural bottleneck, such as inconsistent data formatting from external sources.
• Strategic Alignment: Helps users understand why a specific marketing or research strategy yielded suboptimal results, allowing the agent to suggest future adjustments based on historical performance.

Example Prompts

1. "I attempted to export the quarterly report but the task failed; can you run the failure-analyzer to see why it crashed?"
2. "Why does this automation keep timing out? Perform a 5 Whys analysis on the recent task history."
3. "Summarize all recent technical failures from the last week and suggest a long-term preventive strategy for each."

Tips & Limitations

To get the best results, ensure your task descriptions are descriptive, as the analysis quality depends heavily on the accuracy of the 'expected' vs 'actual' input data. Note that this skill is reactive; while it provides preventive warnings for future tasks, it requires at least one initial occurrence of a failure to establish a baseline. Avoid clearing the /moltworker/data/learning/ directory unless you intend to reset the agent's learned experiences.