terminal-bench-loop

Automate bounded Terminal-Bench test loops with diagnostics and human approval.

Updated Feb 9, 2026
One-click install
npx skills add https://github.com/Chasonnnn/AgentCompany --skill terminal-bench-loop-chasonnnn
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: terminal-bench-loop
Source: https://github.com/Chasonnnn/AgentCompany/tree/main/skills/terminal-bench-loop
Command: npx skills add https://github.com/Chasonnnn/AgentCompany --skill terminal-bench-loop-chasonnnn

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill automates the process of running Terminal-Bench problems repeatedly within Paperclip, enabling efficient, bounded, human-in-the-loop improvement cycles to identify and fix issues systematically.

Core Features & Use Cases

  • Automated Loop Management: Initiates, manages, and diagnoses multiple Terminal-Bench iterations to pass smoke tests reliably.
  • Bounded Runs & Diagnostics: Ensures each test stays within predefined iteration and time limits, providing precise stop reasons and analysis.
  • Use Case: When a Terminal-Bench task stalls or fails intermittently, use this Skill to drive automated reruns, diagnose stop points, and coordinate product fixes with human approval.

Quick Start

Use this Skill to run a Terminal-Bench problem in a bounded loop, diagnose failures, and coordinate fixes with the team.

Frequently Asked Questions about terminal-bench-loop

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

FAQPage Schema
How do I automate Terminal-Bench test loops with bounded iterations and diagnostics?

Automate Terminal-Bench test loops by configuring the system to manage multiple iterations with predefined time and iteration limits, providing precise stop reasons and diagnostic analysis for each failure point.

What is the best way to resolve intermittent Terminal-Bench task stalls during continuous integration?

Resolve intermittent Terminal-Bench task stalls by driving automated reruns within a bounded loop, diagnosing stop points, and coordinating product fixes through a human approval process.

Why does my Terminal-Bench loop need human approval for issue resolution?

Terminal-Bench loops need human approval to coordinate product fixes systematically, ensuring that real blockers are recognized and that each iteration produces productive work before proceeding.

Can I run Terminal-Bench diagnostic loops without maintaining loop invariants?

No, running Terminal-Bench diagnostic loops requires maintaining all loop invariants, including bounded iteration counts, productive work validation, and real blocker recognition, to ensure efficient issue resolution.

Does Terminal-Bench loop automation work for continuous integration testing scenarios?

Yes, Terminal-Bench loop automation is suitable for continuous integration and iterative problem solving scenarios, providing bounded runs and diagnostics to pass smoke tests reliably.

What are the limitations of using bounded loops for Terminal-Bench problem resolution?

Bounded loops for Terminal-Bench problem resolution are limited by predefined iteration and time limits, stopping execution when real blockers are recognized to prevent unproductive cycles.