review
Use when reviewing a PR/MR by URL or the local working-tree diff — runs multi-perspective parallel agents (bug scanner, CLAUDE.md compliance, test coverage, plan alignment, git history, sibling MR/epic coherence) with confidence scoring and optional draft inline comments on the MR. Triggers for `/devflow:review` or any "review this PR/MR" request.
72
88%
Does it follow best practices?
Impact
—
No eval scenarios have been run
Advisory
Suggest reviewing before use
Quality
Discovery
100%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 excellent skill description that clearly communicates what the skill does (multi-perspective PR/MR review with parallel agents covering specific analysis dimensions) and when to use it (explicit triggers including a slash command and natural language patterns). The description is information-dense without being verbose, and its specificity about the review perspectives (bug scanning, test coverage, plan alignment, etc.) makes it highly distinguishable from generic code review skills.
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | Lists multiple specific concrete actions: multi-perspective parallel agents including bug scanning, CLAUDE.md compliance, test coverage, plan alignment, git history, sibling MR/epic coherence, confidence scoring, and optional draft inline comments. Very detailed and concrete. | 3 / 3 |
Completeness | Clearly answers both 'what' (runs multi-perspective parallel agents with confidence scoring and inline comments) and 'when' (explicit triggers: '/devflow:review', 'review this PR/MR' request, reviewing by URL or local working-tree diff). | 3 / 3 |
Trigger Term Quality | Includes strong natural trigger terms: 'PR', 'MR', 'review this PR/MR', '/devflow:review', 'URL', 'working-tree diff'. These cover common variations users would naturally say when requesting code review. | 3 / 3 |
Distinctiveness Conflict Risk | Highly distinctive with a clear niche — PR/MR review with specific multi-agent analysis perspectives. The explicit slash command '/devflow:review' and the detailed scope (bug scanner, CLAUDE.md compliance, epic coherence) make it very unlikely to conflict with other skills. | 3 / 3 |
Total | 12 / 12 Passed |
Implementation
77%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This is a sophisticated, well-structured skill for multi-perspective code review with strong actionability — concrete commands, specific API calls, clear decision logic, and precise cross-referencing patterns. The workflow is excellently sequenced across 6 phases with validation checkpoints and fallback handling. The main weaknesses are moderate verbosity (some rules are stated twice across sections) and the inability to verify the referenced bundle files (AGENTS.md, TEMPLATES.md) that carry significant responsibility for the skill's completeness.
Suggestions
Deduplicate repeated instructions: the Source-of-Truth Hierarchy, pagination mandate, and 'never use local branch diffs' rule each appear in both their respective phases AND the 'Important rules' section — keep them in one place and reference from the other.
Consider moving the detailed Phase 1e parsing instructions (discussion endpoint specifics, bot-fix-commit detection heuristics) into REFERENCE.md to reduce the main skill's length while keeping the high-level flow scannable.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The skill is quite long (~300+ lines) and includes some redundant explanations (e.g., re-explaining the Source-of-Truth Hierarchy in both Phase 1d and the Important Rules section, repeating pagination instructions in both Phase 1e and Important Rules). However, most content is genuinely instructive and domain-specific — not explaining things Claude already knows. Some tightening is possible but the bulk earns its place. | 2 / 3 |
Actionability | Highly actionable throughout: concrete CLI commands for gh/glab/git, specific API endpoints with pagination parameters, exact regex patterns for task ID extraction, specific MCP tool names (mcp__5ebcd1ed-*__getJiraIssue), and clear conditional logic for mode detection. The code blocks are copy-paste ready and the decision tables are precise. | 3 / 3 |
Workflow Clarity | The 6-phase pipeline is clearly sequenced with explicit decision points (Phase 0 mode detection), parallel execution instructions (Phase 2 agent dispatch), validation/cross-check steps (Phase 3 deduplication and existing-discussion cross-check with 4 status categories), and a user confirmation gate before posting comments (Phase 5). Error recovery is addressed via fallback rules (missing integrations) and the cap on large diffs. | 3 / 3 |
Progressive Disclosure | The skill correctly references sibling files (AGENTS.md, TEMPLATES.md, and a design doc) for agent definitions, scoring rubrics, and output templates — good separation of concerns. However, no bundle files were provided to verify these references exist, and the main SKILL.md itself is quite long with substantial inline detail that could potentially be offloaded. The references are clearly signaled and one-level deep, which is good, but the inline content density pushes this to a 2. | 2 / 3 |
Total | 10 / 12 Passed |
Validation
100%Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.
Validation — 11 / 11 Passed
Validation for skill structure
No warnings or errors.
- Repository
- AndreJorgeLopes/devflow
- Commit
b0b1bb6
- 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.