skills.rest
eugene-belkovich
eugene-belkovichCOMMUNITY
Explore all skills by this author →
GitHub

recurse

Automate the Eval→Reflect→Act loop to diagnose failures and generate fix plans.

Updated Aug 23, 2026
Search Tags:#automation#debugging#reliability#reflection#evals#agent-evaluation
One-click install
npx skills add https://github.com/eugene-belkovich/ai-setup --skill recurse
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: recurse
Source: https://github.com/eugene-belkovich/ai-setup/tree/main/claude/profiles/work/skills/evals/recurse
Command: npx skills add https://github.com/eugene-belkovich/ai-setup --skill recurse

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Automates the Eval→Reflect→Act optimization loop to diagnose failures and generate actionable fix plans.

Core Features & Use Cases

  • End-to-end orchestration of evals with automated reflections
  • Categorized failure analysis and prioritized, actionable fix plans
  • Optional overnight autonomous mode (Ralph Wiggum) for continuous improvement

Quick Start

Run the eval suite and review the generated reflections to apply safe fixes.

Frequently Asked Questions about recurse

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

FAQPage Schema
How do I automate eval cycles to diagnose agent failures?▼

Automate eval cycles by running an Eval→Reflect→Act loop that diagnoses failures and generates actionable fix plans. The suite runs deterministic execution to output audit-ready reflections and structured fixes across test domains.

What is the Eval→Reflect→Act optimization loop?▼

The Eval→Reflect→Act optimization loop is an automated process for diagnosing failures and generating actionable fix plans. It runs eval suites, produces automated reflections on failures, and applies safe auto-fixes to resolve identified issues.

How do I generate structured fix plans from eval suite failures?▼

Generate structured fix plans by running the eval suite to trigger automated reflections. The system performs categorized failure analysis and outputs prioritized, actionable fix plans limited to context_mismatch and eval_structural categories.

Can I run continuous overnight eval cycles for autonomous debugging?▼

Yes, you can run continuous overnight eval cycles for autonomous debugging using the optional overnight mode. This mode enables continuous improvement by automating the eval, reflection, and safe auto-fix handling loop without manual intervention.

What types of eval failures are safe for automated fixing?▼

Safe automated fixing is limited to context_mismatch and eval_structural failure categories. The system restricts auto-fix handling to these specific issues to ensure deterministic execution and audit-ready outputs without causing unintended side effects.

Related Skills

Explore More Software Engineering Skills→
rivaldiekaptrrrrivaldiekaptrrr

eval-harness

Define and run eval suites to validate AI task performance.

Community
Advanced
adhi-jpadhi-jp

skill-quality

Classify skill contract failures from eval results and edit SKILL.md definitions.

Community
Advanced
danmestasdanmestas
4

orchestrator-mode

Coordinate subharness fleets for distributed AI workflows with deterministic routing.

Community
Advanced
MatthieuszMatthieusz

diagnosing-bugs

Diagnose and fix software bugs through a structured six-phase process.

Community
Advanced