eagle3-triage

Triage a failed EAGLE3 pipeline run. Identifies which step failed (data synthesis, hidden state dump, training, or benchmark), diagnoses root cause from logs, and suggests fixes. Use when user reports an EAGLE3 pipeline failure or asks why a specific step failed. Also helps debug new model support issues.

Quality

88%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Low

Low-risk findings worth noting

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 executable commands, clear validation checkpoints, and dense diagnostic tables, but it is a single large file with some redundant 'How it works' prose and no progressive disclosure into reference files.

Suggestions

Move the per-step error-pattern tables into reference files (e.g. references/task_1_failures.md) and keep SKILL.md as an overview that links to them, improving progressive_disclosure.

Trim the 'How it works' paragraphs where they restate the Pipeline Overview table, or fold the overview into the per-step sections to remove duplication.

In the fix columns, avoid restating the root cause verbatim (e.g. 'Batch size too large' → 'Reduce batch size'); let the Root cause column carry that so the Fix column holds only the actionable change.

Dimension	Reasoning	Score
Conciseness	The error/root-cause/fix tables are token-efficient, but the per-task 'How it works' paragraphs partially duplicate the Pipeline Overview table and some fixes restate the root cause already named in an adjacent column, so not every token earns its place.	2 / 3
Actionability	Provides fully executable, copy-paste-ready commands (re-run with task skip flags, exact env vars like TIKTOKEN_RS_CACHE_DIR, exact flag positions, and a find/tail log-fetching snippet) rather than pseudocode.	3 / 3
Workflow Clarity	A clear Step 0→5 sequence with an explicit validation checkpoint (find the first task with a non-zero exit code), per-step error→fix feedback loops, a new-model checklist, and re-run skip commands for error recovery.	3 / 3
Progressive Disclosure	Well-sectioned but monolithic: all per-step error catalogs are inline in a single ~8.7KB SKILL.md with no references/ or scripts/ bundle files, and the skill exceeds the 50-line threshold for the section-organization exemption to apply.	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, uses natural trigger phrasing, answers both what and when with an explicit 'Use when' clause, and occupies a distinct niche. It is written in third person with no fluff or over-claims.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions — identifies the failed step (naming all four pipeline steps), diagnoses root cause from logs, and suggests fixes — matching the 'lists multiple specific concrete actions' anchor.	3 / 3
Completeness	Clearly answers both 'what' (identify/diagnose/suggest fixes) and 'when' with an explicit 'Use when...' trigger clause, so it does not hit the cap at 2 for missing trigger guidance.	3 / 3
Trigger Term Quality	The 'Use when user reports an EAGLE3 pipeline failure or asks why a specific step failed' clause uses natural user phrasings rather than jargon, giving good coverage of terms users would actually say.	3 / 3
Distinctiveness Conflict Risk	The EAGLE3 offline pipeline niche with named steps (data synthesis, hidden state dump, training, benchmark) is highly specific and unlikely to trigger for unrelated skills.	3 / 3
	Total	12 / 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
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	15 / 16 Passed

Repository: NVIDIA/Model-Optimizer
Path: .agents/skills/eagle3-triage/SKILL.md
Commit: 87c9f8c

Reviewed: 15 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.