Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Render clean blueprint-style SVG diagrams from JSON specs. Use when users ask to draw, sketch, or diagram a request flow, neural net, transformer block, system architecture, state machine, data pipeline, or any node-and-edge technical visual they want as an SVG for docs, READMEs, posts, or slides.
.claude/skills/davila7-diagrammer/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-06 | ✗→✓ | ▲ Improved | -77% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -79% | 0% |
| case-22 | ✗→✓ | ▲ Improved | -62% | 0% |
| case-11 | ✓→✗ | ▼ Worse | -88% | 0% |
| case-12 | ✓→✗ | ▼ Worse | -89% | 0% |
Use diagrammer to turn a small JSON spec into a clean SVG diagram. It is useful when the user wants a precise technical diagram without opening a design tool.
The renderer must be installed locally:
bashpipx install diagrammer
bashdiagrammer path/to/spec.json > path/to/diagram.svg
json{ "nodes": [ {"id": "client", "type": "box", "label": "client"}, {"id": "api", "type": "box", "label": "api"}, {"id": "db", "type": "database", "label": "postgres"} ], "edges": [ {"from": "client", "to": "api", "label": "request"}, {"from": "api", "to": "db", "label": "query"} ] }
Built-in node types: box, circle, text, database, stack, group, note, and custom.
Useful optional fields:
direction: "LR" or "TB"router: "straight" or "ortho"label on edgesstyle: "solid" or "dashed"weight: "thin" or "thick"For the full reference, run:
bashdiagrammer prompt
Use diagrammer when the output should be a checked-in SVG artifact, especially for:
Prefer Mermaid when the user specifically asks for Mermaid syntax or wants diagrams rendered by a Markdown platform. Prefer Excalidraw when the user wants editable hand-drawn canvas files.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-11 | pass→fail | 26,852 | 4,049 | -85% | 1 | 1 | 0% | 6,173 | 710 | -88% | 0 | 0 | — |
case-01 | fail→fail | 25,320 | 5,729 | -77% | 1 | 1 | 0% | 6,191 | 1,170 | -81% | 0 | 0 | — |
case-02 | fail→fail | 27,620 | 6,467 | -77% | 1 | 1 | 0% | 6,182 | 906 | -85% | 0 | 0 | — |
case-03 | pass→pass | 8,163 | 7,100 | -13% | 1 | 1 | 0% | 1,468 | 1,727 | +18% | 0 | 0 | — |
case-04 | pass→pass | 25,705 | 27,857 | +8% | 1 | 1 | 0% | 6,184 | 6,652 | +8% | 0 | 0 | — |
case-05 | pass→pass | 7,114 | 8,591 | +21% | 1 | 1 | 0% | 1,466 | 2,320 | +58% | 0 | 0 | — |
case-06 | fail→pass | 26,864 | 4,629 | -83% | 1 | 1 | 0% | 6,184 | 1,448 | -77% | 0 | 0 | — |
case-07 | fail→fail | 29,337 | 6,230 | -79% | 1 | 1 | 0% | 6,181 | 1,032 | -83% | 0 | 0 | — |
case-08 | fail→fail | 32,296 | 5,927 | -82% | 1 | 1 | 0% | 6,172 | 958 | -84% | 0 | 0 | — |
case-09 | fail→fail | 28,969 | 5,741 | -80% | 1 | 1 | 0% | 6,174 | 821 | -87% | 0 | 0 | — |
case-10 | fail→fail | 18,050 | 7,198 | -60% | 1 | 1 | 0% | 4,393 | 934 | -79% | 0 | 0 | — |
case-12 | pass→fail | 28,196 | 3,476 | -88% | 1 | 1 | 0% | 6,173 | 678 | -89% | 0 | 0 | — |
case-13 | fail→pass | 16,545 | 1,334 | -92% | 1 | 1 | 0% | 2,987 | 637 | -79% | 0 | 0 | — |
case-14 | fail→fail | 27,551 | 5,148 | -81% | 1 | 1 | 0% | 6,167 | 753 | -88% | 0 | 0 | — |
case-15 | fail→fail | 19,863 | 5,580 | -72% | 1 | 1 | 0% | 4,878 | 705 | -86% | 0 | 0 | — |
case-16 | fail→fail | 27,211 | 5,280 | -81% | 1 | 1 | 0% | 6,167 | 857 | -86% | 0 | 0 | — |
case-17 | fail→fail | 25,217 | 4,856 | -81% | 1 | 1 | 0% | 6,174 | 616 | -90% | 0 | 0 | — |
case-18 | fail→fail | 28,565 | 3,563 | -88% | 1 | 1 | 0% | 6,169 | 589 | -90% | 0 | 0 | — |
case-19 | fail→fail | 29,091 | 8,220 | -72% | 1 | 1 | 0% | 6,176 | 1,136 | -82% | 0 | 0 | — |
case-20 | fail→fail | 25,145 | 6,138 | -76% | 1 | 1 | 0% | 6,179 | 672 | -89% | 0 | 0 | — |
case-21 | fail→fail | 32,953 | 4,717 | -86% | 1 | 1 | 0% | 6,164 | 665 | -89% | 0 | 0 | — |
case-22 | fail→pass | 9,684 | 1,588 | -84% | 1 | 1 | 0% | 1,680 | 638 | -62% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 6 counted toward the lift figure. The other 16 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +5 percentage points is the difference between those two pass rates over the 6 comparable cases. 3 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.