skills.rest
Kjdragan
KjdraganCOMMUNITY
Explore all skills by this author →
GitHub

run-evaluation

Analyzes Universal Agent run logs and traces to identify errors and improvements.

1|1|Updated Dec 21, 2025
Search Tags:#performance#post-mortem#log analysis#evaluate#agent run#trace analysis
One-click install
npx skills add https://github.com/Kjdragan/universal_agent --skill run-evaluation
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: run-evaluation
Source: https://github.com/Kjdragan/universal_agent/tree/main/.agent/workflows/skills/run-evaluation
Command: npx skills add https://github.com/Kjdragan/universal_agent --skill run-evaluation

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill performs a post-mortem analysis of the latest Universal Agent run by inspecting run.log, session outputs, and Logfire traces to identify errors, bottlenecks, and opportunities for improvement.

Core Features & Use Cases

  • Automated post-mortem analysis: Analyze run.log, session outputs, and Logfire traces to surface actionable insights.
  • Problem identification: Detect anomalies, exceptions, and deviations from a happy path.
  • Recommendations: Generate concrete improvement suggestions for future runs, tuning, and debugging.

Quick Start

Run the evaluation workflow after an agent run completes using the steps described in the Skill guide.

Frequently Asked Questions about run-evaluation

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I perform a post-mortem analysis on an agent run?▼

You can perform a post-mortem analysis by inspecting run.log, session outputs, and Logfire traces to identify anomalies, exceptions, and deviations from a happy path, then generating a structured evaluation report.

How does log analysis identify agent performance bottlenecks?▼

Log analysis identifies bottlenecks by inspecting run.log and Logfire traces to detect anomalies, exceptions, and deviations from a happy path, surfacing actionable insights for tuning and iterative improvement.

Do I need access to the agent workspace to evaluate run logs?▼

Yes, you need access to the agent workspace containing AGENT_RUN_WORKSPACES/session_*/run.log and corresponding Logfire traces to evaluate run logs and produce a structured evaluation report.

What's the best way to debug errors in Logfire traces after an agent session?▼

The best way to debug errors is to analyze Logfire traces and session outputs to detect anomalies and exceptions, generating concrete improvement suggestions for future runs and debugging.

Can I generate improvement recommendations from session outputs automatically?▼

Yes, you can automatically generate improvement recommendations by analyzing session outputs and run logs to surface actionable insights and concrete suggestions for future runs and tuning.

Related Skills

Explore More Data & Analytics Skills→
massgenmassgen
1.1k

massgen-log-analyzer

Run MassGen experiments and generate ANALYSIS_REPORT.md from traces.

Official
Advanced
Chai-UnChai-Un

fix

Fix lint errors and formatting issues using yarn, prettier, and linc.

Community
Intermediate
d0whc3rd0whc3r

error-detective

Analyze logs and stack traces to identify root causes across distributed systems.

Community
Advanced
JaganproJaganpro
423

sf-debug

Diagnose Salesforce Apex debug logs to identify governor-limit violations and root causes.

Community
Advanced