Harness Doctor

Overview

The Harness Doctor acts as a critical quality gatekeeper, monitoring the operational stability of an entire agent ecosystem. Its primary function is to analyze failures from the last 24 hours and track patterns across all connected agents. Instead of just diagnosing problems, it proactively files concrete, actionable issues in the development tracker for recurring or high-impact bugs.

Capabilities

• Failure Pattern Clustering: Groups raw failure logs into distinct, repeatable error patterns.
• Threshold Monitoring: Identifies patterns that have occurred 3+ times in the last 7 days OR 2+ times in the last 24 hours.
• Actionable Issue Filing: Drafts detailed Forge issues containing specific file paths and required code diffs (old block vs. new block).
• Issue Management: Prevents duplicate issue filing by checking for existing open reports on the same pattern.
• Daily Reporting: Posts a summary digest detailing pattern counts, issues filed, and updated watch-list status to a routine ticket.

Example Use Cases

• Proactive Bug Fixing: If three different agents fail due to an identical serialization error across multiple modules, the Doctor files one comprehensive issue pointing to the root library fix.
• System Stability Reporting: At the end of each cycle, it provides a concise report to the Forge COO summarizing system health and required developer attention points.
• Closing Feedback Loops: It ensures that identified issues are not merely diagnosed but are formally logged with concrete steps for resolution, moving beyond simple advisory notes.