Execute commands with confidence. The SelfHealing task automatically detects execution errors, analyzes logs, and applies fixes in real-time—turning fragile scripts into resilient workflows.
Automatically analyzes stderr output and exit codes to propose and apply patches without human intervention.
Execute commands in specific working directories relative to the project root, ensuring context-aware execution.
Chain multiple commands together. If one fails, the system attempts to heal it before proceeding to the next.
Optional "Human-in-the-loop" mode allows you to review and approve generated patches before application.
Generates clear diff outputs showing exactly what code was changed to resolve the execution error.
Produces detailed Markdown, HTML, and PDF transcripts of the entire execution and repair session.
Configure a command sequence and watch the agent attempt to fix simulated errors.
// No patches applied yet.
Status: Pending
Automatically fix linting errors or dependency mismatches during build processes.
Run setup scripts that might fail due to missing packages, and let the agent install them.
Iteratively run tests and fix syntax errors when upgrading language versions.