aris-analyze-results

Analyze ML experiment results, compute statistics, generate comparison tables and insights. Use when user says "analyze results", "compare", or needs to interpret experimental data.

Quality

81%

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

The canonical home for this skill is analyze-results in wanshuiyin/Auto-claude-code-research-in-sleep

Quality

Content

80%Weight 40%Scale 1-5

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

The body is a concise, well-structured instruction skill that gives specific analytical guidance and a clean five-step workflow. Its weakness is the absence of validation/verification feedback loops in a batch analysis context, which caps workflow clarity, plus a few steps that remain abstract rather than executable.

Suggestions

Add an explicit validation checkpoint in the workflow (e.g. after building the comparison table, verify all runs are accounted for and baselines are tagged before computing deltas) with a fix-and-retry loop.

Make abstract steps executable by specifying concrete commands or snippets (e.g. a jq/pandas snippet to load results JSON and compute mean/std, or a specific column layout for the comparison table).

Define the output table format concretely (columns and a one-line example row) so the 'Raw data table' deliverable is unambiguous.

Dimension	Reasoning	Score
Conciseness	The body is a lean ~40 lines with no padding and no explanation of concepts Claude already knows; tight conditional clauses like 'If multiple seeds:' and 'If sweeping a parameter:' earn their tokens, matching the lean-and-efficient 5 anchor.	5 / 5
Actionability	Concrete, specific guidance appears throughout ('Delta vs baseline: always compute relative improvement', 'report mean +/- std', the Observation/Interpretation/Implication/Next step scaffold), but a few steps stay abstract ('Parse JSON results into structured data', 'identify trends') with no executable commands, leaving minor gaps below a fully copy-paste-ready 5.	4 / 5
Workflow Clarity	A clear five-step sequence (Locate → Build Table → Statistical Analysis → Insights → Update Docs) plus an output format is present, but the analysis/batch workflow has no validation or verify-then-proceed feedback loop, so per the batch-operations cap it cannot exceed 3.	3 / 5
Progressive Disclosure	Under 50 lines with no external references needed and well-organized sections (numbered Workflow steps + Output Format); as a simple single-purpose skill this meets the 5 anchor for well-organized structure without file references.	5 / 5
	Total	17 / 20 Passed

Description

83%Weight 40%Scale 1-5

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 strong: it pairs a concrete capability list with an explicit, natural-language trigger clause, fully answering both what and when. Its main limitation is the slightly generic 'compare' trigger and a couple of abstract action verbs that keep it just short of maximal specificity and distinctiveness.

Suggestions

Tighten action verbs by replacing abstract terms like 'analyze' and 'insights' with concrete deliverables (e.g. 'compute mean/std across seeds, produce delta-vs-baseline tables, summarize takeaways').

Narrow the 'compare' trigger to the ML context (e.g. 'compare models', 'compare runs', 'compare hyperparameter sweeps') to reduce overlap with general comparison skills.

Add common synonyms and file/term cues users say (e.g. 'perplexity', 'ablation', 'results.json', 'experiment logs') to broaden natural trigger coverage.

Dimension	Reasoning	Score
Specificity	Quotes 'compute statistics, generate comparison tables and insights' alongside 'Analyze ML experiment results' — lists several specific concrete actions, but 'analyze' and 'insights' stay slightly abstract, leaving minor coverage gaps rather than the fully concrete listing of a 5.	4 / 5
Completeness	Explicitly answers both 'what' (analyze results, compute statistics, generate comparison tables and insights) and 'when' via the concrete trigger clause 'Use when user says "analyze results", "compare", or needs to interpret experimental data', matching the 5 anchor.	5 / 5
Trigger Term Quality	Includes natural user phrases 'analyze results', 'compare', and 'interpret experimental data', giving good keyword coverage; missing common synonyms or extensions (e.g. 'compare models', 'perplexity', 'results summary') that a 5 would cover.	4 / 5
Distinctiveness Conflict Risk	The ML experiment-results framing carves a fairly distinct niche, but the bare 'compare' trigger is broad enough to overlap with general comparison skills, leaving minor overlap risk rather than minimal.	4 / 5
	Total	17 / 20 Passed

Validation

87%

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 — 14 / 16 Passed

Validation for skill structure

Criteria	Description	Result
allowed_tools_field	'allowed-tools' contains unusual tool name(s)	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	14 / 16 Passed

Repository: OpenLAIR/dr-claw
Path: skills/aris-analyze-results/SKILL.md
Commit: 01ef795

Reviewed: ~18 hours 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.