agent-code-review-swarm
Agent skill for code-review-swarm - invoke with $agent-code-review-swarm
34
0%
Does it follow best practices?
Impact
98%
2.64xAverage score across 3 eval scenarios
Advisory
Suggest reviewing before use
Optimize this skill with Tessl
npx tessl skill review --optimize ./.agents/skills/agent-code-review-swarm/SKILL.mdQuality
Discovery
0%Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.
This is an extremely minimal description that fails on all dimensions. It provides no information about what the skill does, when it should be used, or what makes it distinct. It reads more like an invocation instruction than a skill description.
Suggestions
Add concrete actions describing what the skill does, e.g., 'Performs multi-agent code review by analyzing code for bugs, style issues, security vulnerabilities, and performance concerns.'
Add an explicit 'Use when...' clause with natural trigger terms, e.g., 'Use when the user asks for a code review, wants feedback on a pull request, or needs code quality analysis.'
Explain what 'swarm' means in this context to distinguish it from a simple code review skill, e.g., 'Uses multiple specialized review agents to provide comprehensive feedback from different perspectives (security, performance, style).'
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | The description contains no concrete actions whatsoever. It only says 'Agent skill for code-review-swarm' without describing what the skill actually does (e.g., reviewing code, finding bugs, suggesting improvements). | 1 / 3 |
Completeness | Neither 'what does this do' nor 'when should Claude use it' is answered. The description only provides an invocation command ('$agent-code-review-swarm') without explaining functionality or trigger conditions. | 1 / 3 |
Trigger Term Quality | The only potentially relevant term is 'code-review-swarm' which is a technical/internal name, not a natural keyword a user would say. Users would say things like 'review my code', 'code review', 'check for bugs', etc. | 1 / 3 |
Distinctiveness Conflict Risk | The description is so vague that it's unclear what distinguishes this from any other code review tool or skill. The term 'swarm' hints at multi-agent behavior but is not explained, making it impossible to differentiate. | 1 / 3 |
Total | 4 / 12 Passed |
Implementation
0%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This skill is a verbose, non-actionable wishlist masquerading as documentation. It references a fictional CLI tool ('npx ruv-swarm') extensively, making none of the code examples executable or useful. The content is bloated with descriptive JSON objects that list concepts Claude already knows (OWASP categories, SOLID principles, etc.) rather than providing concrete, working instructions for performing code reviews with actual tools.
Suggestions
Replace all 'npx ruv-swarm' references with actual working tools and commands (e.g., real gh CLI commands, actual MCP tool calls like mcp__claude-flow__swarm_init, or concrete scripts that exist in the repo)
Remove the descriptive JSON blocks listing security checks, performance metrics, style checks, and architecture patterns - Claude already knows these concepts and they waste hundreds of tokens
Define a clear, sequential workflow: 1) Get PR diff via gh CLI, 2) Spawn specific agents via MCP tools, 3) Collect results, 4) Validate findings, 5) Post review - with explicit validation checkpoints at each step
Reduce the file to under 100 lines covering the core workflow, and move configuration examples, comment templates, and CI/CD integration into separate referenced files
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Extremely verbose at 400+ lines with massive amounts of redundant content. JSON blocks listing check categories (security checks, performance metrics, style checks, architecture patterns) are purely descriptive and add no actionable value. Multiple sections repeat similar patterns (review-init commands with different flags). Claude already knows what SQL injection and XSS are. | 1 / 3 |
Actionability | Nearly all code examples reference a fictional tool 'npx ruv-swarm' with invented CLI flags that don't correspond to any real tool. The commands are not executable - they're aspirational pseudocode dressed up as bash. The JavaScript 'review agent' objects are just JSON data structures listing concepts, not executable code. The GitHub Actions workflow references non-existent actions. | 1 / 3 |
Workflow Clarity | Despite being a multi-step process (initialize swarm → run agents → collect results → post comments), there are no clear validation checkpoints or error recovery steps. The workflow jumps between disconnected code blocks without explaining how they connect. No feedback loops for when reviews fail or agents produce incorrect results. The actual orchestration between agents is never defined. | 1 / 3 |
Progressive Disclosure | Monolithic wall of text with no meaningful separation of concerns. Everything from configuration to monitoring to examples is dumped into one massive file. The two references at the bottom (swarm-pr.md, workflow-automation.md) are mentioned but the main file contains far too much inline content that should be split out. Sections like 'Review Agents' with their JSON descriptor blocks could easily be separate reference files. | 1 / 3 |
Total | 4 / 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 (543 lines); consider splitting into references/ and linking | Warning |
Total | 10 / 11 Passed | |
- Repository
- ruvnet/claude-flow
- Commit
f547cec
- 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.