agent-refinement
Agent skill for refinement - invoke with $agent-refinement
43
13%
Does it follow best practices?
Impact
96%
1.23xAverage score across 3 eval scenarios
Passed
No known issues
Optimize this skill with Tessl
npx tessl skill review --optimize ./.agents/skills/agent-refinement/SKILL.mdQuality
Discovery
0%Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.
This is an extremely weak description that provides almost no useful information for skill selection. It fails on every dimension: no concrete actions, no natural trigger terms, no 'when to use' guidance, and no distinctiveness. It reads more like a label than a functional description.
Suggestions
Define what 'refinement' means concretely — list specific actions like 'iterates on drafts, improves code quality, refines prose for clarity' or whatever the skill actually does.
Add an explicit 'Use when...' clause with natural trigger terms users would say, e.g., 'Use when the user asks to improve, polish, iterate on, or refine existing content or code.'
Specify the domain or input types this skill operates on (e.g., code, text, designs) to distinguish it from other editing or improvement skills.
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | The description contains no concrete actions whatsoever. 'Refinement' is abstract and undefined — it does not specify what is being refined, how, or what outputs are produced. | 1 / 3 |
Completeness | The description fails to answer both 'what does this do' and 'when should Claude use it.' There is no explanation of capabilities and no 'Use when...' clause or equivalent trigger guidance. | 1 / 3 |
Trigger Term Quality | The only potentially relevant keyword is 'refinement,' which is not a natural term users would typically say. The description includes an invocation command ('$agent-refinement') but no natural language trigger terms. | 1 / 3 |
Distinctiveness Conflict Risk | 'Refinement' is extremely generic and could apply to virtually any editing, improving, or iterating task. This would conflict with many other skills that involve any form of content improvement. | 1 / 3 |
Total | 4 / 12 Passed |
Implementation
27%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This skill reads like a general software engineering tutorial rather than a focused agent instruction set. It spends hundreds of lines teaching Claude concepts it already knows (TDD, circuit breakers, error hierarchies, cyclomatic complexity) with generic examples unrelated to any specific project. The content lacks a clear, actionable workflow for what the refinement agent should actually do when invoked, and everything is crammed into one monolithic file.
Suggestions
Reduce content to under 80 lines by removing explanations of concepts Claude already knows (TDD, circuit breakers, error hierarchies, complexity) and focus on the specific refinement workflow steps and decision criteria.
Add a clear, sequenced workflow with validation checkpoints: e.g., 1) Analyze current test coverage, 2) Identify untested paths, 3) Write tests, 4) Run and verify green, 5) Refactor, 6) Re-run tests, 7) Check coverage thresholds.
Split detailed patterns (error handling, performance optimization, circuit breakers) into separate reference files and link to them from a concise overview.
Define concrete entry/exit criteria for the refinement phase—what triggers it, what conditions must be met to consider it complete, and how to report results.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Extremely verbose at ~400+ lines. Explains basic concepts Claude already knows (TDD red/green/refactor, what circuit breakers are, what error hierarchies are, cyclomatic complexity). The lengthy code examples are illustrative rather than project-specific, essentially teaching general software engineering patterns that Claude already understands. | 1 / 3 |
Actionability | The code examples are concrete and executable TypeScript, which is good. However, they are generic teaching examples (authentication service, user permissions) rather than actionable instructions for a specific refinement workflow. There's no guidance on how to actually invoke this skill on a real codebase or what specific steps to take when called. | 2 / 3 |
Workflow Clarity | The TDD phases (Red/Green/Refactor) provide a sequence, but there are no explicit validation checkpoints or feedback loops for the overall refinement process. The skill doesn't define when refinement is 'done,' how to verify improvements, or what to do if metrics aren't met. The numbered sections under 'Performance Refinement' and 'Error Handling Refinement' are more like a catalog of patterns than a clear workflow. | 2 / 3 |
Progressive Disclosure | This is a monolithic wall of text with no references to external files. All content—TDD examples, performance optimization, error handling, circuit breakers, quality metrics, complexity analysis—is inlined in a single massive document. Much of this could be split into separate reference files with a concise overview in the main skill. | 1 / 3 |
Total | 6 / 12 Passed |
Validation
90%Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.
Validation — 10 / 11 Passed
Validation for skill structure
| Criteria | Description | Result |
|---|---|---|
skill_md_line_count | SKILL.md is long (530 lines); consider splitting into references/ and linking | Warning |
Total | 10 / 11 Passed | |
- Repository
- ruvnet/claude-flow
- Commit
322b2ae
- Reviewed
Table of Contents
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.