agentleFS
NanfordNanford/Nanford-skills/.agents/skills/architecture-diagram/SKILL.md

architecture-diagram

Skill18 starsChanged 5 months agoMITView on GitHub

Design polished architecture diagrams, system overviews, platform maps, solution blueprints, technical landscape slides, and layered capability views. Use when creating or revising an architecture visual in HTML, SVG, slides, images, or other visual formats. Default to a connector-free composition with no arrows or relationship lines; communicate structure through containment, aligned layers, grouping, spacing, labels, and restrained color.

48 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use competitor-news-monitor, blocked-page-recovery and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • Samin12Skill

    competitor-news-monitor

    Watch named companies for material news; cited digests.

    Also in 6 projects that use this

  • nobodyohm-webSkill

    blocked-page-recovery

    Recover blocked/paywalled/WAF'd pages via fallbacks.

    Also in 6 projects that use this

  • Samin12Skill

    document-to-action-items

    Extract cited obligations, deadlines, tasks from documents.

    Also in 6 projects that use this

  • VercelSkill

    agent-desktop-linux

    Linux platform guide for the agent-desktop CLI. Lists per command group what works on X11 and on Wayland, how to check the AT-SPI accessibility bus, the first-contact gotchas (app names, Chromium and Electron flags, headed focus, blocked key combos), and where to find lifecycle, screenshot, clipboard, tray, notification and troubleshooting details. Triggers on: "linux desktop automation", "AT-SPI tree", "automate GNOME app", "agent-desktop on Linux", "accessibility bus not available", "empty supported_surfaces", "wayland screenshot consent", or any Linux GUI interaction task.

    Also in 3 projects that use this

  • VercelSkill

    agent-desktop-macos

    macOS platform guide for the agent-desktop CLI. Use for desktop automation on macOS: granting Accessibility and Screen Recording permission (TCC), what headed mode does on macOS (physical-first click, type and scroll), surfaces (menu, menubar, sheet, popover, alert, dock, menu-bar-extras), menu bar apps and system panels with no window, Notification Center, Chromium and Electron apps including launch --cdp, and fixing PERM_DENIED, empty accessibility trees, ACTION_FAILED and APP_UNRESPONSIVE.

    Also in 3 projects that use this

  • michaellam26Skill

    sdlc-review

    Trigger stage-specific reviews for an SDLC project and save results to reviews/

    Also in 5 projects that use this

  • +2 more from vercel-labs
  • +1 more from Samin12

The file

Show its rules and text

We split this skill into the rules it gives an agent

39 rules. 0 are common, 39 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 45 edited versions we read keep these.

  • 39

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Create architecture diagrams that read through hierarchy and spatial organization rather than a web of connectors.

    line 8
  2. Distinctive:

    Do not draw arrows.

    line 12
  3. Distinctive:

    Do not draw connector lines, leader lines, flow lines, dependency lines, dotted paths, or decorative lines that could be mistaken for …

    line 13
  4. Distinctive:

    Do not use arrow characters or directional chevrons between components.

    line 14
  5. Distinctive:

    Do not reserve empty corridors for imaginary connectors.

    line 15
  6. Distinctive:

    Treat any visible connector as a defect unless the user explicitly requires connections.

    line 16
  7. Distinctive:

    If the user explicitly requests connections, first offer a connector-free alternative.

    line 17
  8. Distinctive:

    Use these visual signals, in priority order:

    line 21
  9. Distinctive:

    Containment: Put related components inside a clearly titled parent region.

    line 23
  10. Distinctive:

    Columns or swimlanes: Use columns for peer domains, channels, products, environments, or actors.

    line 25
  11. Distinctive:

    Proximity: Keep related items close and separate unrelated groups with generous whitespace.

    line 26
  12. Distinctive:

    Numbering: When order matters, use small step numbers or stage labels instead of arrows.

    line 29
  13. Distinctive:

    Never imply a false relationship merely to make the composition symmetrical.

    line 32
  14. Distinctive:

    Use for systems with clear tiers.

    line 40
  15. Distinctive:

    Use for business domains or product families.

    line 53
  16. Distinctive:

    Use only when one genuine platform or product is central.

    line 57
  17. Distinctive:

    Use for ownership or deployment boundaries.

    line 61
  18. Distinctive:

    Do not mix several composition patterns on one canvas.

    line 63
  19. Distinctive:

    Remove duplicate concepts and collapse low-value implementation details.

    line 68
  20. Distinctive:

    Place parent regions before child cards.

    line 73
  21. Distinctive:

    Verify the reading order without relying on connectors.

    line 74
  22. Distinctive:

    Add concise labels and optional one-line subtitles.

    line 75
  23. Distinctive:

    When the source material describes many point-to-point interactions, abstract them into a shared layer such as “Integration & APIs,” “Event …

    line 78
  24. Distinctive:

    Use one clear title and, when useful, one short subtitle.

    line 82
  25. Distinctive:

    Use a consistent spacing unit and generous outer margins.

    line 83
  26. Distinctive:

    Use a 12-column or similarly regular grid.

    line 84
  27. Distinctive:

    Keep peer cards equal in width, height, padding, corner radius, and heading position.

    line 85
  28. Distinctive:

    Use 3–6 layers and no more than 7 top-level groups.

    line 86
  29. Distinctive:

    Use one neutral canvas, one primary accent, and no more than 3–5 restrained domain tints.

    line 87
  30. Distinctive:

    Ensure text has strong contrast.

    line 88
  31. Distinctive:

    Use subtle borders or soft shadows, not both heavily.

    line 89
  32. Distinctive:

    Keep icons optional, small, and stylistically consistent.

    line 90
  33. Distinctive:

    Prefer short noun phrases of 1–4 words.

    line 91
  34. Distinctive:

    Place shared services in one clearly labeled full-width region adjacent to the consumers.

    line 118
  35. Distinctive:

    If exact network, sequence, or dependency semantics are the main subject and omitting connectors would make the diagram inaccurate:

    line 122
  36. Distinctive:

    Prefer a matrix, numbered interaction table, or paired “source / relationship / target” list beside the architecture view.

    line 125
  37. Distinctive:

    If connectors are still explicitly required, use only direct orthogonal connectors, one semantic style, no crossings, and a separate …

    line 126
  38. Distinctive:

    Before delivering, verify every item:

    line 130
  39. Distinctive:

    If any check fails, revise the hierarchy or grouping before adding visual decoration.

    line 143

“Common” is measured: kept by at least 6 in 10 of the 45 edited versions of this file we could read. “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

Architecture Diagram

Create architecture diagrams that read through hierarchy and spatial organization rather than a web of connectors.

Non-negotiable default

  • Do not draw arrows.
  • Do not draw connector lines, leader lines, flow lines, dependency lines, dotted paths, or decorative lines that could be mistaken for relationships.
  • Do not use arrow characters or directional chevrons between components.
  • Do not reserve empty corridors for imaginary connectors.
  • Treat any visible connector as a defect unless the user explicitly requires connections.
  • If the user explicitly requests connections, first offer a connector-free alternative. Add only the minimum requested connections when they remain necessary.

Communicate relationships without lines

Use these visual signals, in priority order:

  1. Containment: Put related components inside a clearly titled parent region.
  2. Vertical hierarchy: Arrange broad concerns at the top and foundational concerns at the bottom.
  3. Columns or swimlanes: Use columns for peer domains, channels, products, environments, or actors.
  4. Proximity: Keep related items close and separate unrelated groups with generous whitespace.
  5. Alignment: Align peers to a shared grid and keep their dimensions consistent.
  6. Color families: Give each domain one restrained tint; use color consistently, not decoratively.
  7. Numbering: When order matters, use small step numbers or stage labels instead of arrows.
  8. Short captions: State the relationship in a group subtitle when spatial placement alone is insufficient.

Never imply a false relationship merely to make the composition symmetrical.

Choose one composition

Select the simplest pattern that matches the content:

Layered stack

Use for systems with clear tiers. Create 3–6 full-width horizontal bands. Place peer cards inside each band.

Typical order:

  • Experience / channels
  • Business capabilities / applications
  • Shared services / integration
  • Data / intelligence
  • Platform / infrastructure
  • Governance / security as a side rail or quiet cross-cutting band

Domain columns

Use for business domains or product families. Create 3–5 equal columns, each containing a short vertical stack. Put shared foundations in one full-width band below the columns.

Core and surrounding domains

Use only when one genuine platform or product is central. Place the core in a large central panel and arrange surrounding domains as separate nearby cards. Use distance, headings, and color—not radial lines—to show association.

Nested platform map

Use for ownership or deployment boundaries. Nest components inside 2–3 levels of containers. Avoid excessive nesting and never place tiny text inside deeply nested boxes.

Do not mix several composition patterns on one canvas.

Build the hierarchy

  1. Extract the title, scope, audiences, domains, components, and true parent-child relationships.
  2. Remove duplicate concepts and collapse low-value implementation details.
  3. Limit the top level to 3–7 groups.
  4. Limit each group to 2–6 child components.
  5. Assign every component exactly one primary visual home. Repeat an item only when the repeated label is explicitly marked as shared.
  6. Choose a single composition pattern.
  7. Place parent regions before child cards.
  8. Verify the reading order without relying on connectors.
  9. Add concise labels and optional one-line subtitles.
  10. Apply restrained styling only after the structure is correct.

When the source material describes many point-to-point interactions, abstract them into a shared layer such as “Integration & APIs,” “Event Backbone,” or “Data Exchange” rather than drawing one line per interaction.

Visual system

  • Use one clear title and, when useful, one short subtitle.
  • Use a consistent spacing unit and generous outer margins.
  • Use a 12-column or similarly regular grid.
  • Keep peer cards equal in width, height, padding, corner radius, and heading position.
  • Use 3–6 layers and no more than 7 top-level groups.
  • Use one neutral canvas, one primary accent, and no more than 3–5 restrained domain tints.
  • Ensure text has strong contrast; do not rely on color alone to encode meaning.
  • Use subtle borders or soft shadows, not both heavily.
  • Keep icons optional, small, and stylistically consistent. Never use icons as substitutes for labels.
  • Prefer short noun phrases of 1–4 words. Keep descriptions to one line when possible.
  • Make the visual hierarchy obvious at thumbnail size: title, group headings, then component labels.

Avoid

  • Spiderweb, flowchart, circuit-board, or subway-map layouts.
  • Crossing lines, curved arrows, bidirectional arrows, and arrows entering container borders.
  • Decorative strokes between cards.
  • Tiny labels floating in whitespace.
  • More than three levels of visible nesting.
  • Unequal gaps, arbitrary card sizes, or misaligned baselines.
  • Excessive gradients, glow, glass effects, 3D perspective, or ornamental backgrounds.
  • A separate box for every noun in the source.
  • Legends that are needed only because the color system is too complex.

Special cases

Process or data flow

Convert the flow into numbered stages arranged left-to-right or top-to-bottom. Use large stage numbers and whitespace between stages. Do not connect stages.

Cross-cutting concerns

Show security, governance, observability, and operations as a side rail, top cap, or bottom foundation. Do not draw a line from each concern to every component.

Shared services

Place shared services in one clearly labeled full-width region adjacent to the consumers. Do not duplicate them inside every domain.

Required connection semantics

If exact network, sequence, or dependency semantics are the main subject and omitting connectors would make the diagram inaccurate:

  1. Tell the user that the request conflicts with the connector-free default.
  2. Prefer a matrix, numbered interaction table, or paired “source / relationship / target” list beside the architecture view.
  3. If connectors are still explicitly required, use only direct orthogonal connectors, one semantic style, no crossings, and a separate detailed view rather than cluttering the overview.

Quality gate

Before delivering, verify every item:

  • There are no arrows, arrowheads, connector lines, relationship strokes, or directional chevrons.
  • The diagram remains understandable in grayscale.
  • The title, groups, and components form three visibly distinct text levels.
  • Every component has one unambiguous parent or region.
  • Peers align to a consistent grid and use consistent dimensions.
  • Whitespace clearly separates unrelated groups.
  • No group is overcrowded; details have been consolidated or moved to a secondary view.
  • The reading order is obvious without explanation.
  • Cross-cutting concerns use a band or rail rather than repeated links.
  • The final view is accurate, balanced, and legible at its intended display size.

If any check fails, revise the hierarchy or grouping before adding visual decoration.

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.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

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.