Analyzes agent execution traces to detect and score workflow inefficiencies.
npx skills add https://github.com/ApiliumCode/mayros --skill workflow-insights
Please help me install this Agent Skill. Skill: workflow-insights Source: https://github.com/ApiliumCode/mayros/tree/main/skills/official/workflow-insights Command: npx skills add https://github.com/ApiliumCode/mayros --skill workflow-insights
This Skill identifies inefficiencies and anti-patterns within agent workflows, helping to optimize performance and reduce resource waste.
Analyze the current agent's workflow traces to identify any inefficiencies.
High-intent search queries and answers about installing and using this skill.
You can detect agent workflow inefficiencies by analyzing execution traces with a weighted system of 7 anti-pattern detectors. This process identifies issues like repeated failures, long delegation chains, unused tools, and error cascades to pinpoint exactly where pipelines break down.
A workflow health score is a 0-100 metric calculated by evaluating execution traces against 7 anti-pattern detectors. It assesses delegation, error patterns, resource usage, and tool utilization to provide an automated health assessment for multi-step agent pipelines.
Debug a slow agent by running an execution trace analysis to identify anti-patterns like repeated failures and long delegation chains. The analysis scores workflow inefficiencies, highlighting exact breakdown points in tool utilization and resource usage.
Yes, you can audit agent performance and resource waste using workflow analysis scripts and references without external dependencies. The evaluation applies anti-pattern detectors directly to execution traces to score delegation and error cascades.
Workflow analysis for multi-step agent pipelines is limited to detecting known inefficiencies like redundant queries and timeouts. It relies entirely on execution trace data, meaning it cannot diagnose external network issues or logic flaws outside the recorded agent workflow.
mandubianAggregate cross-session traces to distill learnings and per-agent performance signals.
lazygophersGuide JVM tuning, garbage collection, and performance analysis for Java 21+ applications.
Forexgod21Map delegation chains and trust boundaries in multi-agent AI systems.
alsk1992Monitors system health and provides alerts via configurable thresholds.
