agent-neural-network

Agent skill for neural-network - invoke with $agent-neural-network

1.60x

Quality

41%

Does it follow best practices?

Impact

96%

1.60x

Average score across 3 eval scenarios

Securityby

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./.agents/skills/agent-neural-network/SKILL.md

Quality

Content

46%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 persona-heavy agent definition: the MCP toolkit examples are the strongest asset, but roughly half the file is role-play padding and ML trivia Claude already knows. The embedded duplicate frontmatter block (name: flow-nexus-neural) inside the body is structurally confusing for a SKILL.md.

Suggestions

Cut the architecture explainer, quality standards, and advanced capabilities lists — Claude already knows what LSTMs and GANs are — and keep only the Flow Nexus-specific tool parameters and workflows.

Make the six-step workflow actionable by attaching concrete tool calls or commands to each step and adding a validation checkpoint (e.g. check training status/metrics before promoting a model to inference).

Remove or fix the embedded second YAML frontmatter block (lines 6-10), which duplicates name/description inside the markdown body and can break frontmatter parsing.

Dimension	Reasoning	Score
Conciseness	Multiple padded sections explain concepts Claude already knows ("Feedforward: Classic dense networks for classification and regression", generic "Quality standards" and "Advanced capabilities" lists), plus a boilerplate persona paragraph. This is noticeably verbose rather than severely verbose — the toolkit code block is genuinely useful — so anchor 2 fits better than 1.	2 / 5
Actionability	The toolkit block gives concrete MCP tool calls (neural_train with a full config object, neural_cluster_init, neural_predict), but placeholders like "model_id" and "user_id" and abstract step descriptions ("Problem Analysis: Understand the ML task") leave it incomplete. It sits between anchor 3's incomplete guidance and anchor 4's mostly-executable guidance; the placeholder IDs and missing output/response details pull it to 3.	3 / 5
Workflow Clarity	A six-step ML workflow is listed in order, but each step is a one-line abstraction with no commands, no validation checkpoints, and no error-recovery feedback loops (e.g. what to do when training fails or metrics regress). This matches anchor 3 (sequence present, checkpoints missing) rather than anchor 4.	3 / 5
Progressive Disclosure	No bundle files exist and the body has clear section headers with no dangling or nested references, so navigation is straightforward. Minor gaps — the architecture descriptions, quality standards, and advanced capabilities lists could be trimmed or split — keep it at anchor 4 rather than 5.	4 / 5
	Total	12 / 20 Passed

Description

36%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 a generated placeholder that names the domain but tells Claude nothing about what the skill does or when to use it. It lacks any capability statements, trigger conditions, or differentiators beyond the domain name.

Suggestions

Replace the placeholder with a capability statement drawn from the body, e.g. "Trains, deploys, and manages neural networks via Flow Nexus distributed cloud infrastructure — architecture design, distributed training, inference, and model versioning."

Add an explicit trigger clause: "Use when the user asks to train, tune, deploy, or run inference on a neural network, or mentions Flow Nexus, federated learning, or distributed ML training."

Mention the concrete tool names (neural_train, neural_cluster_init, neural_predict) to distinguish this skill from generic ML or other Flow Nexus agent skills.

Dimension	Reasoning	Score
Specificity	"Agent skill for neural-network - invoke with $agent-neural-network" names the domain but lists no concrete actions (no training, deployment, or inference verbs). It is above anchor 1 because the domain is specifically named rather than pure abstraction, but below anchor 3 because zero concrete capabilities are described.	2 / 5
Completeness	The 'what' is vague ("Agent skill for neural-network") and there is no 'Use when...' trigger clause — "invoke with $agent-neural-network" is an invocation hint, not usage guidance, so completeness is capped at 3 and the vague what with no real when matches anchor 2. Not anchor 1 because the domain is at least stated.	2 / 5
Trigger Term Quality	"neural-network" is a natural keyword a user would say, but the description misses common variations and synonyms such as train, deploy, model, ML, or deep learning. This matches 'some relevant keywords but missing common variations' rather than anchor 4's good coverage.	3 / 5
Distinctiveness Conflict Risk	The neural-network domain is somewhat specific, but the description is generic boilerplate ("Agent skill for X - invoke with $agent-X") that would look identical to sibling agent skills; no Flow Nexus, training, or tooling terms differentiate it. It overlaps with similar ML-related skills but is not entirely generic, so anchor 3 fits best.	3 / 5
	Total	10 / 20 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: ruvnet/ruflo
Path: .agents/skills/agent-neural-network/SKILL.md
Commit: b14c79e

Reviewed: 3 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.