ov-debug

Troubleshooting all sorts of failures, crashes, exceptions and errors using debug capabilities. Analyze accuracy, performance, model compilation, or memory issues. Dump tensors and intermediate blobs. Serialize and visualize IRs, execution graphs. Enable verbose, logging. Profile execution. Compare layer outputs. Inspect, trace or dump transformations. Identify executed operations, nodes, primitives, kernels.

Quality

62%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./.claude/skills/ov-debug/SKILL.md

Quality

Content

65%

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

The body is admirably concise and follows a sound progressive-disclosure design (overview + routing table + one-level references), but its actionability and workflow are weakened because the referenced component files are missing and the routing steps remain generic with no validation checkpoints.

Suggestions

Provide the referenced component files (components/debug-intel-cpu-plugin.md, debug-intel-gpu-plugin.md, debug-transformations.md) so the progressive-disclosure path actually resolves.

Replace the generic 'Follow the instructions and recommendations' step with concrete, executable debug actions or at minimum a validation checkpoint (e.g. how to confirm the identified component is the right one).

Add a verification step in the workflow, such as 'After debugging, re-run the failing case to confirm the issue is resolved before reporting.'

Dimension	Reasoning	Score
Conciseness	The body is lean and efficient — a concise prerequisites list, a routing table, and three short steps — with no padding or explanation of concepts Claude already knows, matching the 'every token earns its place' anchor.	3 / 3
Actionability	It gives one concrete, executable item ('-DENABLE_DEBUG_CAPS=ON' with 'check CMakeCache.txt in the build dir') and a routing table, but the actual debugging guidance is deferred to component reference files that are not present, and the Steps ('Follow the instructions and recommendations') are vague.	2 / 3
Workflow Clarity	A three-step sequence is listed (match symptom → load reference → follow instructions), but the steps are generic and lack any validation or feedback checkpoints for confirming the chosen component or debug result.	2 / 3
Progressive Disclosure	The SKILL.md is well-structured as an overview with a clearly signaled one-level-deep routing table pointing to component files, but those referenced files (components/debug-intel-cpu-plugin.md, debug-intel-gpu-plugin.md, debug-transformations.md) do not exist in the bundle, so the disclosure path is broken.	2 / 3
	Total	9 / 12 Passed

Description

60%

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 rich in specific, concrete debugging actions and clearly targets the OpenVINO debug niche, but it lacks an explicit 'Use when...' trigger clause and leans on technical jargon over natural user phrasings. These gaps cap completeness and trigger-term quality at the mid level.

Suggestions

Add an explicit 'Use when...' clause naming natural trigger phrases users would say, e.g. 'Use when an OpenVINO model crashes, produces wrong results, runs slowly, or fails to compile.'

Soften the leading 'Troubleshooting all sorts of failures' to a more distinctive, OpenVINO-scoped statement to reduce overlap with generic debugging skills.

Pair the technical jargon (IRs, blobs, primitives) with the everyday terms users actually say (wrong results, slow inference, memory leak) to broaden natural trigger coverage.

Dimension	Reasoning	Score
Specificity	The description lists many concrete actions such as 'Dump tensors and intermediate blobs', 'Serialize and visualize IRs, execution graphs', 'Profile execution', and 'Compare layer outputs' — matching the 'lists multiple specific concrete actions' anchor.	3 / 3
Completeness	The 'what' is thoroughly covered, but there is no 'Use when...' clause or equivalent explicit trigger guidance, so 'when' is only implied — the rubric guideline caps completeness at 2 in that case.	2 / 3
Trigger Term Quality	It includes natural terms users would say ('crashes, exceptions and errors', 'accuracy, performance', 'memory issues') but mixes in heavy technical jargon ('IRs', 'intermediate blobs', 'primitives, kernels') and omits common phrasings, fitting the 'some relevant keywords but missing common variations' anchor rather than full coverage.	2 / 3
Distinctiveness Conflict Risk	The specialized OpenVINO vocabulary (IRs, execution graphs, transformations, primitives, kernels) signals a clear niche, but the generic opening 'Troubleshooting all sorts of failures' and absence of explicit trigger context leave some overlap risk with general debugging skills.	2 / 3
	Total	9 / 12 Passed

Validation

93%

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 — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
relative_links	Relative link issues: 3 missing	Warning

	Total	15 / 16 Passed

Repository: openvinotoolkit/openvino
Path: .claude/skills/ov-debug/SKILL.md
Commit: 4c20980

Reviewed: 4 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.