debug-inference

Debug why inference.local, direct external inference, or supervisor-only system inference is failing. Use when the user cannot reach a local model server, has provider base URL issues, sees inference verification failures, hits protocol mismatches, or needs to diagnose inference on local vs remote gateways. Trigger keywords - debug inference, inference.local, system inference, sandbox-system, local inference, ollama, vllm, sglang, trtllm, NIM, inference failing, model server unreachable, failed to verify inference endpoint, host.openshell.internal.

Quality

88%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Medium

Suggest reviewing before use

Quality

Content

77%Weight 40%Scale 1-3

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The body is highly actionable with a well-sequenced, validated diagnostic workflow, but it is longer than necessary due to repetition and keeps all detail inline rather than splitting some material into reference files.

Suggestions

Consolidate the 'Common Failure Patterns' table with the inline workflow steps to remove duplicated cause/fix guidance and reduce length.

Move the long 'Fix: Local Host Inference Timeouts (Firewall)' section and the 'Full Diagnostic Dump' into separate reference files linked from the main body to improve progressive disclosure.

Trim repeated interpretation notes (e.g., host.openshell.internal topology warnings appear in both Step 0 and the failure table) to a single canonical statement.

Dimension	Reasoning	Score
Conciseness	The ~380-line body avoids concept-explanation fluff and assumes competence, but repeats material across the step-by-step workflow, the 'Common Failure Patterns' table, and the firewall/full-diagnostic-dump sections that could be tightened.	2 / 3
Actionability	Provides fully executable, copy-paste-ready commands (openshell, curl, docker exec) alongside specific error strings and concrete fix examples, matching the 'fully executable code/commands' anchor.	3 / 3
Workflow Clarity	Sequences a clear Step 0–7 diagnostic order with an early-stop rule and explicit validation/feedback loops ('Verify the Problem', 'Verify the Fix', 'If It Still Fails').	3 / 3
Progressive Disclosure	No bundle files exist and the skill is a single ~380-line monolith; the firewall fix and full diagnostic dump sections are inline content that could be split into separate reference files, so it is structured but not optimally partitioned.	2 / 3
	Total	10 / 12 Passed

Description

100%Weight 40%Scale 1-3

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description is specific, complete, and distinctive with an explicit 'Use when' clause and a broad set of natural trigger keywords. It is concise and uses proper third-person voice, with no over-claims or fluff.

Dimension	Reasoning	Score
Specificity	Names three concrete inference paths ("inference.local, direct external inference, or supervisor-only system inference") and specific failure modes rather than vague actions, matching the 'lists multiple specific concrete actions' anchor.	3 / 3
Completeness	Clearly answers both 'what' (debug why inference is failing) and 'when' via the explicit 'Use when the user cannot reach a local model server, has provider base URL issues...' clause.	3 / 3
Trigger Term Quality	Includes an explicit, natural keyword list ("debug inference, inference.local, ... ollama, vllm, sglang, trtllm, NIM, inference failing, model server unreachable") giving good coverage of terms a user would actually say.	3 / 3
Distinctiveness Conflict Risk	Occupies a clear niche (OpenShell inference debugging) with distinct trigger tokens like 'inference.local', 'sandbox-system', and 'host.openshell.internal', making overlap with unrelated skills unlikely; voice is third person.	3 / 3
	Total	12 / 12 Passed

Validation

100%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository: NVIDIA/OpenShell
Path: .agents/skills/debug-inference/SKILL.md
Commit: deced87

Reviewed: 6 days ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.