What is the difference between Full Review and Quick Status mode?

Full Review performs a deep 6-phase analysis including cost auditing and pattern detection, while Quick Status provides a fast, read-only summary of current progress, costs, and ETA without generating a full report.

What happens after the review is complete?

The skill proposes follow-up actions such as archiving results, cleaning the runtime state, or moving tasks to a queue for future planning based on the analysis findings.

How does Arc Review detect complexity mismatches?

The skill compares task performance against benchmarks. For example, it flags Haiku tasks that involve more than one file or exceed 8 turns, suggesting they be moved to a more capable model like Sonnet.

Can I check the status of a run while it is still active?

Yes. Using the --quick flag or status phrases will check the .heartbeat file. If it's less than 30 seconds old, the skill will notify you that the engine is active and provide a real-time progress update.

Arc Run Analyzer & Reviewer

Name: Arc Run Analyzer & Reviewer
Author: itsdevcoffee

byitsdevcoffee

•

Analytics & Monitoring

Analyzes Maximus Loop execution performance, tracks progress, and provides cost-to-complexity insights for AI-driven workflows.

Arc Review is a specialized skill designed to monitor and evaluate the performance of Maximus Loop executions within the Claude Code environment. It provides developers with a structured, 6-phase analysis of automated tasks, identifying failure patterns, detecting complexity mismatches, and offering detailed cost estimations based on model usage. Whether you need a comprehensive post-run report with actionable recommendations or a quick status update on a live execution, this skill ensures your AI agent workflows are running efficiently, reliably, and within budget.

Key Features

01Actionable Reporting: Generates structured findings including success rates, top expensive tasks, and specific optimization steps.

02Anomaly Detection: Identifies timeouts, repeated failures, and tasks that require splitting due to high complexity.

036-Phase Full Review: Sequential analysis from metadata tracking to follow-up proposals.

04Cost & Complexity Auditing: Automatically flags tasks that exceed pricing benchmarks for Haiku, Sonnet, or Opus models.

05Quick Status Mode: Provides a lightweight, read-only summary of progress, ETA, and current health for active runs.

061 GitHub stars

Use Cases

01Troubleshooting automated code changes by analyzing failure patterns and reviewing git diffs post-execution.

02Monitoring long-running Maximus Loop agent executions for real-time progress and completion ETAs.

03Auditing AI model expenditures to identify specific tasks that caused budget overruns.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add itsdevcoffee/devcoffee-agent-skills review

For use in Claude.ai and ChatGPT

Download Skill