Conciseness

Dense with operational specifics Claude does not know (direct Mermaid→PNG '-e' export is broken, the '#R' URL format vs '#create=' rewrite, zlib-vs-pako byte mismatch, --disable-gpu for headless GPU failures) with almost no concept tutorials. Minor over-explanations keep it from 5: explaining that backticks in bash are command substitution, and the export command appearing in both 'The pipeline' and the 'Export command' section.

Actionability

Fully executable, copy-paste-ready guidance throughout: exact CLI invocations ('drawio -x -f xml -o diagram.drawio diagram.mmd'), the Node zlib URL one-liner, a complete PowerShell CLI-detection script, a full flag table, per-platform open commands, and concrete WSL2/Windows '.url' workarounds. The examples cover all common output paths (drawio/png/svg/pdf/url).

Workflow Clarity

The pipeline is explicitly sequenced (Author → .drawio → Deliver → Open) with an explicit validation checkpoint ('Run this right after writing XML to a .drawio file, before any layout, export or URL step' — the repair pass) and real feedback loops: a troubleshooting table mapping each failure to cause, fix, and retry (e.g. 'Retry the same command with --disable-gpu added') plus fallback chains when the CLI is absent.

Progressive Disclosure

External references are well-signaled with explicit fetch conditions ('Only when a diagram needs layers, tags, metadata and placeholders… also fetch the full XML reference') and scripts/find-shapes.js is a real, correctly-referenced bundle file — better signaling than anchor 3's example. However, ~150 lines of Windows/WSL2-specific CLI-location chains and .url-file workarounds are inlined in SKILL.md, making it a ~400-line manual rather than an overview pointing to detailed materials. Not 4 because that inline bulk is more than a minor organization gap; not 2 because the structure and signaling are good.

Specificity

Lists several concrete actions and artifacts ('create, generate, draw, or design a diagram, flowchart, architecture diagram, ER diagram…', 'diagram export to PNG/SVG/PDF'), but what the skill actually produces (native .drawio files via Mermaid or XML) is only implied through trigger phrasing rather than stated as a capability. Not 5 because the 'what' is never directly asserted; not 3 because the action and format coverage is broad and concrete.

Completeness

The 'when' is answered exhaustively ('Always use when user asks to create, generate, draw, or design…'), but the 'what' is conveyed only through the trigger verbs ('asks to create… a diagram', 'diagram export to PNG/SVG/PDF') — the skill's actual behavior is never stated as its own clause. Not 5 because anchor 5 requires both what and when explicitly stated; not 3 because both elements are present, just unevenly.

Trigger Term Quality

Comprehensive natural-language coverage including synonyms and file extensions: every common diagram type ('flowchart, architecture diagram, ER diagram, sequence diagram, class diagram, network diagram, mockup, wireframe, UI sketch'), product names ('draw.io, drawio, drawoi', '.drawio files') — even a common misspelling — and export formats ('PNG/SVG/PDF'). Matches the comprehensive-synonyms-and-extensions anchor exactly.

Distinctiveness Conflict Risk

Clear draw.io niche with product-specific triggers ('.drawio files', 'drawoi') and active conflict management against the MCP tool ('Exception: … where the draw.io MCP server's create_diagram tool is available, call that tool instead'). Not 5 because generic triggers like 'diagram', 'flowchart' and 'mockup' would also fire for closely related diagramming/charting skills; not 3 because the niche and disambiguation are far stronger than 'somewhat specific'.