Content
92%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
An exceptionally actionable, well-structured skill: executable dual-language (Python + Cypher) examples throughout, a validated multi-step workflow with confirmation gates and error-recovery guidance, and clean offloading of catalog detail to two well-signaled reference files. The only cost is token weight — duplicated sentences and inline time-sensitive version tables keep it just short of maximal leanness.
Suggestions
Remove the verbatim-duplicated Leiden sentence ('Leiden is defined for directed and undirected graphs...') — keep it once, under community detection — and consolidate the repeated 'AGA Sessions → neo4j-aura-graph-analytics-skill' disclaimers into the When NOT to Use section.
Move the client/server version-compatibility tables into a dedicated 'Version compatibility' or old-patterns section (or a reference file) so time-sensitive version churn does not sit in the Pre-flight working path.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense tables and executable code with no explanations of concepts Claude already knows, but 'Leiden is defined for directed and undirected graphs. Project undirected relationships...' appears verbatim twice, the AGA-sessions redirect repeats several times, and time-sensitive version-compat tables ('2.3 | >= 2.13, < 2.28, < 2026.11') sit in Pre-flight rather than a deprecated/old-patterns section (the 1.x mapping is properly segregated). Efficient overall with minor trimmable instances — a 4, not a 5. | 4 / 5 |
Actionability | Fully executable, copy-paste-ready Python and Cypher for every operation (projection, all algorithms, estimation, cleanup), plus pip install commands, a 12-row error→cause→fix table, and an MCP tool mapping. Coverage of common cases is complete. | 5 / 5 |
Workflow Clarity | The numbered Full Workflow has explicit validation checkpoints (plugin verification via gds.server_version(), memory estimation before large work, stream-to-verify before mutate/write), the Common Errors table provides feedback loops for recovery, write-cypher operations require user confirmation, and a final checklist reinforces the sequence — matching the top anchor. | 5 / 5 |
Progressive Disclosure | SKILL.md is a working overview; the bulk algorithm catalog and projection deep-dive are split into references/algorithms.md and references/graph-projection.md (verified real files), one level deep and clearly signaled in the References section with per-file content descriptions. Easy navigation, appropriate split. | 5 / 5 |
Total | 19 / 20 Passed |