How does this skill help when an agent is stuck in a loop?

It detects circular reasoning patterns—where the same steps are repeated without progress—and provides a specific fix recommendation to help the agent break the loop.

What is 'Instruction Drift' in this context?

Instruction drift occurs when an agent's actions begin to diverge from the original task description over time; this skill flags that divergence so you can re-align the agent.

Does this skill automatically fix the code?

No, it focuses on diagnosis. It writes a detailed reasoning-diagnosis.md file that tells you exactly what went wrong and how to fix the instructions or environment.

What logs does the Reasoning Trace Optimizer analyze?

The skill specifically analyzes the trace-light.log file located in the .ai/traces/ directory to identify patterns in the agent's reasoning and tool usage.

Reasoning Trace Optimizer

Name: Reasoning Trace Optimizer
Author: hjemmesidekongen

byhjemmesidekongen

0•

分析与监控

Diagnoses and optimizes agent reasoning patterns by analyzing session logs for structural failure patterns and bottlenecks.

The Reasoning Trace Optimizer is a diagnostic tool designed to improve AI agent performance by scanning the trace-light.log for common failure modes. It identifies critical issues such as context degradation, tool confusion, and circular reasoning that often cause agent sessions to stall or drift from their original goals. By producing a structured diagnosis report with specific fix recommendations, this skill enables developers to debug underperforming sessions and refine their AI workflows for higher reliability and better task completion rates.

主要功能

01Generates actionable diagnosis reports in markdown with specific recommendations for every issue found.

02Helps prevent compute waste by flagging circular reasoning loops early in the development cycle.

03Deep analysis of session logs to surface hidden bottlenecks and repetitive tool usage.

040 GitHub stars

05Identifies 'premature conclusions' where agents claim completion without proper verification.

06Automated detection of 6 core reasoning failure patterns including goal abandonment and instruction drift.

使用场景

01Identifying the root cause of 'tool confusion' when an agent uses the wrong tools for a task.

02Auditing complex workflows to extract patterns for improving future AI system instructions.

03Debugging an agent session that stalled or failed to reach its intended goal.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add hjemmesidekongen/ai reasoning-trace-optimizer

For use in Claude.ai and ChatGPT

主要功能

01Generates actionable diagnosis reports in markdown with specific recommendations for every issue found.

02Helps prevent compute waste by flagging circular reasoning loops early in the development cycle.

03Deep analysis of session logs to surface hidden bottlenecks and repetitive tool usage.

040 GitHub stars

05Identifies 'premature conclusions' where agents claim completion without proper verification.

06Automated detection of 6 core reasoning failure patterns including goal abandonment and instruction drift.

使用场景

01Identifying the root cause of 'tool confusion' when an agent uses the wrong tools for a task.

02Auditing complex workflows to extract patterns for improving future AI system instructions.

03Debugging an agent session that stalled or failed to reach its intended goal.