agent-sona-learning-optimizer

Agent skill for sona-learning-optimizer - invoke with $agent-sona-learning-optimizer

Quality

34%

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 ./.agents/skills/agent-sona-learning-optimizer/SKILL.md

Quality

Content

36%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 capability brochure rather than an operational skill: it describes what the SONA agent is and quotes benchmark numbers, but gives only two loosely-specified hook commands and no workflow for actually applying the learning loop. It also carries structural noise — a duplicated second YAML frontmatter block and references to files that do not exist in the bundle. The most valuable fix is converting the marketing sections into a concrete pre-task/retrieve/execute/post-task workflow with real parameter examples.

Suggestions

Remove the embedded second YAML block and the benchmark metrics section; replace them with a numbered workflow, e.g. 1. pre-task hook with a concrete description string, 2. retrieve k=3 patterns via the documented command, 3. execute task applying the top pattern, 4. post-task hook recording outcome and success criteria.

Make the hook examples concrete and complete: show real values for --description and explain how --task-id is obtained (e.g. echoed by the pre-task hook), since the current "$TASK"/"$ID" placeholders are unexecutable as written.

Fix or remove the dangling references: either ship 'docs/RUVECTOR_SONA_INTEGRATION.md' in the bundle (e.g. under references/) and link it clearly, or drop the reference along with the '@ruvector$sona@0.1.1' package line, which is metadata rather than guidance. Also correct the corrupted '$' characters (ops$sec, decisions$sec, ruvector$sona) that render as invalid tokens.

Dimension	Reasoning	Score
Conciseness	Capabilities are listed twice — a stray YAML 'capabilities:' block is embedded in the body and then repeated almost verbatim in 'Core Capabilities' — and the 'Performance Characteristics' section is marketing metrics ('+55% quality improvement', '2211 ops$sec', '0.447ms per-vector') that guide no action, constituting several padded sections per anchor 2. It is above 1 because there is no tutorial-style explanation of concepts Claude already knows.	2 / 5
Actionability	The two 'npx claude-flow@alpha hooks' commands are concrete and executable in form, but they use unexplained placeholders ('--description "$TASK"', '--task-id "$ID"' with no indication of where the task id originates) and the remaining sections only describe capabilities ('Automatic model selection', 'Retrieve k=3 similar patterns') without instructing how to perform them. This matches anchor 3: some concrete guidance but incomplete with missing key details.	3 / 5
Workflow Clarity	Only an implicit pre-task then post-task sequence exists, with no instruction on when in a task lifecycle to run the hooks, how to combine them with pattern retrieval or routing, and no validation or error-recovery checkpoint for the recorded outcome. This matches anchor 2 ('rough sequence present but many gaps; validation absent') rather than 3, which would require clearly listed steps.	2 / 5
Progressive Disclosure	Section headers exist and the body is short, but the sole pointer — 'Integration Guide: docs/RUVECTOR_SONA_INTEGRATION.md' — references a file that does not exist in the bundle (no references/, scripts/, or docs/ files ship at all), and the package reference '@ruvector$sona@0.1.1' is not a navigable path. Structure is present but the one reference is a dangling pointer, matching anchor 3.	3 / 5
	Total	10 / 20 Passed

Description

32%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 an auto-generated wrapper that names the skill but communicates nothing about what it does or when to use it. It lacks any capability statement, natural trigger terms, and a 'Use when...' clause. Its only strength is the distinct invocation token that limits conflict risk.

Suggestions

Replace the boilerplate with a capability summary drawn from the body, e.g. 'Continuously improves task quality via SONA adaptive learning: retrieves k=3 similar past patterns, applies LoRA fine-tuning with EWC++ memory preservation, and routes requests across LLMs by cost/quality.'

Add an explicit trigger clause, e.g. 'Use when the user mentions SONA, agent self-optimization, learning from past task outcomes, or LoRA/EWC fine-tuning.'

Include natural synonyms (agent learning, continual learning, pattern reuse, model routing) so the description surfaces when users phrase the need in everyday language rather than the exact skill slug.

Dimension	Reasoning	Score
Specificity	The description names the skill ('sona-learning-optimizer') but states no capabilities — the only action offered is 'invoke with $agent-sona-learning-optimizer', which is a generic invocation hint rather than a concrete action, matching the anchor 'Names the domain but actions are minimal or generic'. It is above anchor 1 because the domain is named specifically rather than in purely abstract terms.	2 / 5
Completeness	The 'what' is extremely vague ('Agent skill for sona-learning-optimizer' says nothing the skill does) and the 'when' is entirely absent with no 'Use when...' clause, which caps completeness at 3. It scores 2 rather than 1 only because the skill name provides a vestige of a 'what'; it is not 3 because no capability or usage condition is stated at all.	2 / 5
Trigger Term Quality	The only terms present are the technical slugs 'sona-learning-optimizer' and '$agent-sona-learning-optimizer'; there are no natural phrases a user would say (e.g. learning, fine-tuning, LoRA, agent optimization). This sits between anchor 1 (jargon only) and anchor 3 (some relevant everyday keywords), closer to 2 since the skill name itself could act as a keyword but no common variations exist.	2 / 5
Distinctiveness Conflict Risk	The '$agent-sona-learning-optimizer' invocation token is a highly specific, unlikely-to-collide trigger, giving a mostly distinct niche. It is not 5 because the generic 'Agent skill for...' framing would pattern-match against any similarly auto-wrapped skill description.	4 / 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-sona-learning-optimizer/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.