agentleFS
Marwan-Ahmed-devMarwan-Ahmed-dev/vibecode-pro-max/.claude/skills/vc-tech-graph/SKILL.md

vc:tech-graph

Skill0 starsChanged 3 months agoMITView on GitHub

Use when you need publish-grade SVG or PNG technical diagrams for architecture, flow, sequence, UML, state, or comparison visuals, with preview used afterward for review rather than generation.

We split this skill into the rules it gives an agent

13 rules. About 0 are common, 5 are distinctive, and 8 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 8

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Do use it for architecture diagrams, data flow diagrams, flowcharts, sequence diagrams, state machines, ER diagrams, use-case diagrams, …

    line 18
  2. Very specific:

    Do use it when the output should live under process/general-plans/references/, process/general-plans/reports/visuals/, …

    line 19
  3. Very specific:

    Do use preview after generation when you want explanation, HTML framing, Mermaid review, or visual self-check of the finished artifact.

    line 20
  4. Distinctive:

    Do not use it as a plan or execution workflow.

    line 21
  5. Distinctive:

    Do not use it to replace inline doc diagrams when a lightweight Mermaid or ASCII explanation is enough.

    line 22
  6. Very specific:

    Use tech-graph when the output should become a durable artifact:

    line 46
  7. Very specific:

    Use preview instead when the goal is:

    line 53
  8. Distinctive:

    create inline Mermaid-first discussion material

    line 58
  9. Very specific:

    Prefer .svg as the source artifact.

    line 62
  10. Very specific:

    If the user does not specify an output path, route outputs to the matching the kit process folder rather than ad hoc locations like …

    line 68
  11. Very specific:

    When the diagram type is ambiguous, classify it before drawing.

    line 89
  12. Very specific:

    Use a template from templates/ or fixture structure from fixtures/ when it speeds up clean SVG generation.

    line 96
  13. Distinctive:

    If useful, review the output visually and iterate.

    line 115

≈ “Common” is an estimate: at least 1 in 5 of the 308,232 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Tech Graph

Generate publish-grade technical diagrams as SVG first, then PNG when raster export is needed.

This is a helper skill only.

  • Do use it for architecture diagrams, data flow diagrams, flowcharts, sequence diagrams, state machines, ER diagrams, use-case diagrams, timelines, and comparison matrices that should become durable artifacts.
  • Do use it when the output should live under process/general-plans/references/, process/general-plans/reports/visuals/, process/features/{feature}/references/, or process/features/{feature}/reports/visuals/.
  • Do use preview after generation when you want explanation, HTML framing, Mermaid review, or visual self-check of the finished artifact.
  • Do not use it as a plan or execution workflow.
  • Do not use it to replace inline doc diagrams when a lightweight Mermaid or ASCII explanation is enough.

Capability Surface

The local the kit skill now includes upstream-style assets:

  • 7 visual style references
  • style-to-diagram guidance matrix
  • icon and SVG layout reference material
  • reusable SVG templates
  • JSON regression fixtures
  • template-driven generation script
  • batch style validation script

Load the smallest references that fit the request:

  • references/style-diagram-matrix.md to choose a style and diagram family
  • references/style-1-flat-icon.md through references/style-7-openai.md for exact visual language
  • references/icons.md for semantic shapes and product-icon guidance
  • references/svg-layout-best-practices.md for spacing, routing, label, and export discipline
  • references/tech-graph-workflow.md for the kit-specific routing and proof rules

Routing Matrix

Use tech-graph when the output should become a durable artifact:

  • process/general-plans/references/
  • process/general-plans/reports/visuals/
  • process/features/{feature}/references/
  • process/features/{feature}/reports/visuals/

Use preview instead when the goal is:

  • explain a concept visually
  • review an existing diagram
  • generate HTML slides or recaps
  • create inline Mermaid-first discussion material

Output Policy

Prefer .svg as the source artifact. Generate .png alongside it when:

  • the user wants an image for sharing
  • you want a visual proof artifact
  • you need to inspect the render rather than only the XML

If the user does not specify an output path, route outputs to the matching the kit process folder rather than ad hoc locations like ~/Desktop or /tmp.

Diagram Types

The imported templates and references support these common families directly:

  • architecture
  • data-flow
  • flowchart
  • sequence
  • comparison
  • timeline
  • mind-map
  • agent
  • memory
  • use-case
  • class
  • state-machine
  • er-diagram
  • network-topology

When the diagram type is ambiguous, classify it before drawing. If the request is really an explanation-first visual, switch to preview instead of forcing a publish-grade diagram.

Workflow

  1. Decide if the task needs publish-grade SVG/PNG or just preview.
  2. Classify the diagram type and pick the output destination.
  3. Load the matching style reference. Default to style 1 unless the request implies another visual language.
  4. Use a template from templates/ or fixture structure from fixtures/ when it speeds up clean SVG generation.
  5. Generate the SVG directly or render it with:
python3 .claude/skills/tech-graph/scripts/generate-from-template.py <template-type> <output.svg> '<json-data>'
  1. Validate the result:
bash .claude/skills/tech-graph/scripts/validate-svg.sh <file.svg>
  1. Export PNG when needed:
bash .claude/skills/tech-graph/scripts/generate-diagram.sh -o <file.svg>
  1. If useful, review the output visually and iterate.

Helper Scripts

  • scripts/generate-diagram.sh
    • validates an SVG
    • exports a sibling PNG when rsvg-convert is available
    • warns and succeeds in SVG-only mode when rsvg-convert is missing
  • scripts/generate-from-template.py
    • renders starter diagrams from the imported templates plus JSON structure
    • supports the upstream visual style system and semantic shapes
  • scripts/test-all-styles.sh
    • renders the regression fixtures
    • validates every generated SVG
    • exports PNG proofs when rsvg-convert is present
    • writes batch outputs to a the kit process artifact folder by default
  • scripts/validate-svg.sh
    • checks XML structure and closing tags
    • uses rsvg-convert as an extra validator when available

Dependency Contract

  • Required for SVG validation: shell, Python 3, and basic Unix tools
  • Optional but recommended for PNG export: rsvg-convert

the kit allows degraded SVG-only mode when rsvg-convert is unavailable. The skill should warn clearly instead of failing silently.

Good Trigger Phrases

  • generate architecture diagram
  • draw the system flow
  • make this sequence diagram publishable
  • create a clean SVG for this process
  • turn this architecture into PNG
  • generate a UML state machine
  • build a comparison matrix for these systems

Load references/tech-graph-workflow.md for the the kit-specific routing, proof, and preview-boundary rules.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

Reviews can't be read right now.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.