Content
78%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.
A tight, well-structured overview for a simple code-backed skill: a concrete Quick Start and a correctly placed one-level reference to the implementation script. The remaining gaps are a redundant Purpose line and a Quick Start that is not fully self-contained (undefined project_dir, missing initialize() for loading existing graphs).
Suggestions
Delete or merge the Purpose section; it repeats the opening line ('Build and query knowledge graphs for project understanding') almost verbatim.
Make the Quick Start copy-paste ready: define project_dir (e.g., Path.cwd()), show an async context or note the await requirement, and mention calling initialize() to load an existing graph.
State persistence behavior in one line (e.g., 'Graph persists to .claude/knowledge-graph.json; query(entity, depth=N) traverses relations'), so users don't need to open the script for these key facts.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean with well-chosen code and no padding, but the Purpose line 'Maintains a knowledge graph of project concepts, relationships, and learnings...' substantially duplicates the opening line 'Build and query knowledge graphs for project understanding' and the frontmatter description. That minor redundancy is exactly the level 4 anchor's 'minor instances that could be trimmed', keeping it below level 5's 'every token earns its place'. | 4 / 5 |
Actionability | The Quick Start gives a concrete, mostly executable usage sequence (instantiate, add_entity, add_relation, query) covering the common path. It is not copy-paste ready: 'project_dir' is undefined, top-level 'await' has no async context, and 'initialize()' for loading an existing graph is never shown — minor gaps matching level 4 rather than level 5. | 4 / 5 |
Workflow Clarity | As a simple single-purpose skill, the Quick Start's linear usage pattern is unambiguous for the happy path. However, the workflow omits how to load an existing graph (initialize()), where data persists (.claude/knowledge-graph.json), and the query depth option — minor gaps consistent with level 4 rather than a fully self-contained level 5. | 4 / 5 |
Progressive Disclosure | The body is under 50 lines, single-task, and organized into clear sections (Purpose, Quick Start, API Reference), with a real, clearly signaled one-level-deep reference to scripts/knowledge_graph.py (verified to exist in the bundle) that appropriately holds the bulk implementation. Per the rubric's simple-skill scoring note and the level 5 anchor, this is a clear overview with well-signaled one-level references and easy navigation. | 5 / 5 |
Total | 17 / 20 Passed |