debug-inference

Diagnose OpenShell inference failures and produce corrected gateway, provider, and network configurations.

8.1k|1.2k|Updated Feb 24, 2026
One-click install
npx skills add https://github.com/NVIDIA/OpenShell --skill debug-inference-nvidia
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: debug-inference
Source: https://github.com/NVIDIA/OpenShell/tree/main/.agents/skills/debug-inference
Command: npx skills add https://github.com/NVIDIA/OpenShell --skill debug-inference-nvidia

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Diagnose why OpenShell inference is failing or unreachable, including local model server problems, provider URL issues, verification failures, protocol mismatches, and gateway routing concerns.

Core Features & Use Cases

  • Structured diagnostics: Determine whether inference is managed (inference.local) or external, and identify gateway topology issues.
  • Stepwise remediation: Guide users through checking gateway status, managed inference config, provider records, network reachability, sandbox probes, and remediation commands.
  • Real-world use case: When a user cannot reach a local model server, the skill helps isolate whether the problem is the gateway, provider, or network policy and suggests concrete fixes.

Quick Start

Run the diagnostic workflow to identify and fix the active gateway's inference path.

Frequently Asked Questions about debug-inference

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

FAQPage Schema
How do I debug OpenShell inference failures affecting local endpoints?

To debug OpenShell inference failures, run a structured diagnostic workflow to check gateway status, managed inference configurations, and provider records for inference.local. This process isolates root causes and provides actionable remediation steps to restore endpoint reachability.

Why does my inference.local endpoint verification fail?

Endpoint verification for inference.local fails due to gateway topology issues, misconfigured managed inference settings, or network policy restrictions. Diagnosing the active gateway's inference path and sandbox probing helps identify the specific protocol mismatch or unreachable host causing the failure.

How do I fix a misconfigured OPENAI_BASE_URL in OpenShell?

Fixing a misconfigured OPENAI_BASE_URL requires verifying external provider records and host reachability through sandbox probing. The diagnostic workflow checks network policies and gateway routing to produce a corrected, hot-reloadable configuration for the external endpoint.

Can I diagnose external provider routing issues with this inference debugger?

Yes, you can diagnose external provider routing issues by verifying gateway status and checking provider records against network policies. The debugger isolates whether the failure stems from the gateway, the external provider, or a protocol mismatch.

What is the best way to isolate gateway topology problems during inference debugging?

The best way to isolate gateway topology problems is to determine whether inference is managed internally or external, then stepwise check gateway status and sandbox probes. This structured approach identifies routing concerns and suggests concrete remediation commands.

OpenShell inference not working after network policy changes, what should I check?

When OpenShell inference stops working after network policy changes, check host reachability and sandbox probing results against the active gateway configuration. The debugger identifies blocked paths and generates a hot-reloadable configuration to restore connectivity.