| name | stacks-technical-diagrams |
| description | Create polished, dependency-free architecture, workflow, sequence, data-flow, and lifecycle diagrams for Stacks applications as standalone HTML with inline SVG, dark/light themes, and PNG, JPEG, WebP, and SVG export. Use for system or cloud architecture, security boundaries, network topology, technical workflows, CI/CD, runbooks, API call sequences, request lifecycles, data pipelines, lineage, PII boundaries, state machines, status transitions, or converting Mermaid into a purpose-built diagram. |
| license | MIT |
| metadata | {"source":"tt-a1i/archify","source_version":"2.10.0","source_commit":"73f769d2e97c7fb54b74bebb276c1987db278c3f","based_on":"Cocoon-AI/architecture-diagram-generator (MIT, v1.0)"} |
Technical Diagrams for Stacks
Create professional technical diagrams as self-contained HTML files with inline SVG, a theme toggle, and a built-in image/SVG export menu.
Every diagram ships with a dark/light theme toggle (persists in localStorage, respects prefers-color-scheme), an export menu (copy PNG to clipboard; download PNG/JPEG/WebP rasterized natively at up to 4× resolution; download a dual-theme SVG that follows the embedding host's prefers-color-scheme - ideal for GitHub READMEs), and a CSS-variable color system that keeps both themes consistent.
Attribution
This skill ports the renderer and visual system from Archify 2.10.0 by tt-a1i under the MIT license. Archify is itself a fork and rewrite of Cocoon AI's architecture-diagram-generator 1.0. Preserve LICENSE when redistributing this skill or substantial portions of its source.
Dependency-free contract
Do not install packages or fetch runtime assets. Use Bun and the bundled source only. Standalone validators for all five JSON schemas are checked in, the renderer uses Bun's built-in Node compatibility modules, and generated HTML uses system font fallbacks with no network requests.
Run bun --config=storage/framework/defaults/ai/skills/stacks-technical-diagrams/bunfig.toml --no-env-file storage/framework/defaults/ai/skills/stacks-technical-diagrams/bin/technical-diagrams.mjs doctor from the Stacks project root to verify the skill. Run bun --config=storage/framework/defaults/ai/skills/stacks-technical-diagrams/bunfig.toml --no-env-file storage/framework/defaults/ai/skills/stacks-technical-diagrams/bin/technical-diagrams.mjs demo [output-directory] to generate a ready-to-open example before creating the first custom diagram.
Stacks repository workflow
- Inspect the implementation before drawing. Treat
app/ as overrides and storage/framework/defaults/app/ as fallbacks. Follow registrations in app/Routes.ts, route files, actions, jobs, listeners, middleware, models, config, resources, and framework entry points that matter to the requested view.
- Read the relevant Stacks domain skill before mapping an unfamiliar subsystem. Common companions include
stacks-router, stacks-actions, stacks-models, stacks-database, stacks-jobs, stacks-events, stacks-realtime, and stacks-cloud.
- Draw one question per diagram. For a runtime overview, prefer browser or client -> stx/router -> action or service -> model/query -> database, then add only the external systems and trust boundaries needed for that story.
- Write output to the user's requested path. When no path is given, use
docs/diagrams/<descriptive-name>.html and keep the source JSON beside it.
If Bun cannot run, fall back to architecture mode: copy assets/template.html, hand-place SVG using the design system below, and run the self-review checklist before delivering.
Choosing a Diagram Type
| Type | Use for | How |
|---|
architecture | System components, cloud resources, services, security boundaries, infrastructure | renderers/architecture/render-architecture.mjs + JSON (or hand-place SVG when renderers can't run) |
workflow | Technical flows, approval gates, tool calls, runbooks, CI/CD, incident response | renderers/workflow/render-workflow.mjs + JSON |
sequence | API call chains, request lifecycles, cache fallback, async traces, return paths | renderers/sequence/render-sequence.mjs + JSON |
dataflow | Pipelines, ETL/ELT, PII isolation, lineage, warehouse sync, consumers | renderers/dataflow/render-dataflow.mjs + JSON |
lifecycle | State machines, status transitions, wait states, retries, terminal states | renderers/lifecycle/render-lifecycle.mjs + JSON |
Trigger phrases: "architecture/system/cloud diagram" → architecture (unless clearly process-oriented). "workflow/flow/process/runbook/approval/CI-CD/incident" → workflow. "sequence/interaction/call chain/who calls whom" → sequence. "data flow/pipeline/ETL/lineage/PII/governance" → dataflow. "state/status/lifecycle/state machine/retry/terminal" → lifecycle.
Mermaid as an Input Dialect
When the user pastes Mermaid code, do NOT try to render or parse it mechanically. Read it for structure and lay out from scratch in the matching diagram mode:
| Mermaid | Diagram mode | Mapping |
|---|
flowchart / graph | workflow (or architecture if it's a component map) | subgraph → lane or region boundary; node shape {} (diamond) → decision/security node; --> labels → edge labels (use sparingly); classDef/style → nearest semantic type |
sequenceDiagram | sequence | participant → participants (pick semantic type from the name); ->> → message, -->> → return variant; Note → message note; rect blocks → segments |
stateDiagram | lifecycle | states → states (pick start/active/waiting/success/failure from names); [*] start/end → start type / terminal lane; transition labels → event-like labels |
Drop Mermaid styling; keep only the topology and meaning. You choose grouping, lane order, and what deserves emphasis - that judgment is the product.
Layout principles (read before placing)
The renderer's readability comes from spatial narrative, not from drawing every dependency as an arrow. Before you write coordinates or edge lists, plan one clear story:
- One main path - left → right (architecture) or lane → column (workflow). The reader should trace the happy path without crossing lines.
- Few labeled edges - label only cross-boundary or non-obvious transitions on the main path. Adjacent steps stay unlabeled.
- Short side branches - permissions, storage, bots, CI: connect up or down from the nearest node on the main path. Never route a secondary edge diagonally across unrelated components.
- Cards for detail - policies, tech stack notes, and "also connects to X" belong in summary cards, not as extra arrows.
- Mode fit - process / approval / tool-call stories →
workflow or sequence. Component maps with ≤12 nodes → architecture. If the diagram needs 20+ edges, remove edges until the main path is obvious.
Use the complete examples in examples/ as the starting point for each renderer mode.
When validation fails on label overlap, read the Suggested fix lines (coordinates / labelAt / labelDy) and apply them directly - do not guess offsets blindly.
Renderer Modes (architecture / workflow / sequence / dataflow / lifecycle)
All five modes follow the same loop:
- Read first: the schema (
schemas/<type>.schema.json) and the complete worked example (examples/*.{architecture,workflow,sequence,dataflow,lifecycle}.json) - copy its patterns instead of guessing field shapes.
- Write
<name>.<type>.json.
- Render:
bun --config=storage/framework/defaults/ai/skills/stacks-technical-diagrams/bunfig.toml --no-env-file storage/framework/defaults/ai/skills/stacks-technical-diagrams/bin/technical-diagrams.mjs render <type> <input>.json <output>.html.
- Validate the generated artifact:
bun --config=storage/framework/defaults/ai/skills/stacks-technical-diagrams/bunfig.toml --no-env-file storage/framework/defaults/ai/skills/stacks-technical-diagrams/bin/technical-diagrams.mjs validate <type> <input>.json --json, or check an existing HTML file with bun --config=storage/framework/defaults/ai/skills/stacks-technical-diagrams/bunfig.toml --no-env-file storage/framework/defaults/ai/skills/stacks-technical-diagrams/bin/technical-diagrams.mjs check <output>.html. This catches malformed SVG output, non-finite SVG values, two-point diagonal arrows, and arrows crossing the legend.
- If either step fails, the error names the JSON path or the fix (thresholds, valid ranges, which knob to change). Fix the JSON and re-run; never edit the renderer.
Schema violations exit non-zero with path-prefixed messages like /nodes/3 (id/label: "router") must NOT have additional properties. The renderers additionally fail fast on layout problems: node/state overlap (including cross-lane), labels colliding with nodes or other labels, labels wider than their node, out-of-range columns/rows, too-short edges, workflow edges crossing unrelated nodes, and legends outside the viewBox. CJK text is measured at double width automatically.
Set meta.animation: "trace" only when the user asks for motion or a presentation/demo view. It adds lightweight SVG/CSS trace animation to renderer-marked arrows and nodes, respects prefers-reduced-motion, and leaves the default static output unchanged.
Workflow
{
"schema_version": 1,
"diagram_type": "workflow",
"meta": { "title": "Release Workflow", "subtitle": "PR to production", "output": "release.html" },
"lanes": [ { "id": "dev", "label": "Developer" }, { "id": "ci", "label": "CI" }, { "id": "exceptions", "label": "Exception Handling", "variant":
Layout budget: 6 columns (col 0-5) at fixed x positions [88, 220, 300, 430, 500, 625] - columns 1↔2 and 3↔4 are only 70-80px apart, so default-width (92px) nodes in those adjacent columns of the same lane overlap; skip a column or shrink width. Lane content width is 640px. Omit meta.viewBox - the renderer sizes height to the lane count automatically. Use phases for top-of-diagram story beats, groups to frame parallel work or a branch inside one lane, and lane.variant: "exception" for error/retry/fallback lanes. mainPath is optional but recommended: list the happy-path node ids in order so the renderer can catch missing edges or accidental backward movement. Edge routes: straight, drop (bend between lanes; bias 0-1 picks where), outside-right, return-left, bottom-channel, up-channel, or explicit via points. Keep adjacent-step edges unlabeled; reserve labels for cross-lane transitions, approvals, async traces, and returns.
Sequence
{
"schema_version": 1,
"diagram_type": "sequence",
"meta": { "title": "Cache Miss Request", "subtitle": "auth and cache fallback", "output": "cache-miss.html" },
"participants": [
{ "id": "web", "type": "frontend", "label": "Web App", "sublabel": "React UI" },
{ "id": "api", "type": "backend", "label": "API",
Layout budget: participants sit at x = 62 + index×108, so a 920-wide viewBox fits at most 8. Message y must stay within [160, viewBox_height − 83]; messages that share horizontal space need ≥28px vertical separation; arrows need ≥60px horizontal span. segments[].from/to and activations[].from/to are y pixel coordinates, not participant ids. A taller meta.viewBox (default [920, 760]) buys more timeline room. Keep labels short: "GET /path", "verify JWT", "cache miss", "200 JSON".
Dataflow
{
"schema_version": 1,
"diagram_type": "dataflow",
"meta": { "title": "Product Analytics", "subtitle": "events to consumers", "output": "analytics.html" },
"stages": [ { "label": "Sources" }, { "label": "Ingest" }, { "label": "Store" } ],
"nodes": [
{ "id": "web", "type": "frontend",
Layout budget: 2-5 stages at x = 100 + stage×215; 5 rows (row 0-4) at y [128, 242, 356, 470, 584]; default node 112×58. Default viewBox [940, 720]. Flow labels are mandatory and asset-like ("clickstream", "identity map", "feature vectors"); put sensitivity in classification ("PII touch", "approved only", "non-PII"). Variants: emphasis = primary path, security = PII/policy/consent, dashed = async/batch.
Lifecycle
{
"schema_version": 1,
"diagram_type": "lifecycle",
"meta": { "title": "Agent Run Lifecycle", "subtitle": "states and terminal outcomes", "output": "agent-run.html" },
"lanes": [
{ "id": "main", "label": "Lifecycle phases" },
{ "id": "waiting", "label": "Interruptions" },
{ "id": "terminal", "label": "Terminal exits" }
]
Layout budget - lane ids are semantic and reserved: main is required and maps to the top phase band (cols 0-4); terminal maps to the bottom outcome band (cols 0-2); every other lane id shares the single middle event band (cols 0-2) - separate same-band states with different col or yOffset. Band headers render from your lane labels. Default viewBox [980, 660]. Keep transition labels event-like and sparse ("retry", "timeout", "cancel"); prefer state tags, step numbers, and summary cards over label-heavy arrows. Put terminal states in the terminal lane so endings are unambiguous.
Per-mode deep guidance
Each renderer has a README with its full design language (route presets, semantic types, story guidance): renderers/workflow/README.md, renderers/sequence/README.md, renderers/dataflow/README.md, renderers/lifecycle/README.md. Read the matching one before your first diagram of that mode in a session.
Architecture Mode
Architecture has the same read-schema-then-render loop as the other modes - prefer it. Hand-placed SVG is the fallback for when renderers can't run.
{
"schema_version": 1,
"diagram_type": "architecture",
"meta": { "title": "Sample Web App", "subtitle": "3-tier SaaS on AWS", "output": "web-app.html" },
"components": [
{ "id": "users", "type": "external", "label": "Users", "sublabel": "Browser", "pos": [40, 300] },
{ "id": "api",
Render: bun --config=storage/framework/defaults/ai/skills/stacks-technical-diagrams/bunfig.toml --no-env-file storage/framework/defaults/ai/skills/stacks-technical-diagrams/bin/technical-diagrams.mjs render architecture <input>.json <output>.html.
Free placement - pos: [x, y] is the component's top-left; size: [w, h] defaults to [120, 60]. Unlike typed modes there is no lane/stage grid - asymmetric placement is yours to choose. meta.viewBox is optional (auto-fitted).
Grid placement (#8) - when manual coordinates are painful, set semantic cells instead of doing arithmetic:
{
"layout": { "mode": "grid", "cols": 7, "origin": [40, 100], "gapX": 24, "gapY": 48, "cellW": 120, "cellH": 60 },
"components": [
{ "id": "agents", "type": "frontend", "label": "Agent Hosts", "row": 1, "col": 1 }
pos still wins when present and overrides one cell. This is not auto-layout. Spacing is fixed cell math.
Inspect layout (#9) - after editing JSON, dump computed boxes without opening HTML:
bun --config=storage/framework/defaults/ai/skills/stacks-technical-diagrams/bunfig.toml --no-env-file storage/framework/defaults/ai/skills/stacks-technical-diagrams/bin/technical-diagrams.mjs inspect architecture my.architecture.json
Output includes component rects, boundaries, connection point paths, and label positions.
The renderer does the mechanical work that used to be hand-tuned, so you only choose coordinates and meaning:
- Free coordinates -
pos: [x, y] is the component's top-left; size: [w, h] defaults to [120, 60]. Unlike the typed modes there is no lane/stage grid - asymmetric placement is yours to choose. meta.viewBox is optional (auto-fitted to your components + a legend row).
- Grid placement - optional
layout.mode: "grid" with row/col per component (see above). Not dagre; fixed cell spacing only.
- Boundaries from
wraps - list the component ids a region (dashed amber) or security-group (dashed rose) encloses; the renderer computes the box with correct 30/50 padding automatically. Never hand-arithmetic a boundary again.
- Connections route like edges (
variant, fromSide/toSide, route: straight|orthogonal-h|orthogonal-v|auto, via, labelDx/labelDy/labelAt). For a vertical labeled connection, push the label into the gap with labelDy (the validator will tell you if it lands on a box).
- The renderer auto-emits the two-rect
c-mask pattern, draws arrows before boxes (z-order), builds the legend from the component types you used, and fails fast on component overlap, off-canvas components/boundaries, unknown wraps/connection ids, label-vs-component collisions, and non-finite coordinates - the same reliability the other four modes already had.
Hand-placed fallback (no renderer available)
When Bun cannot run, copy assets/template.html and place SVG by hand. Study the worked diagram inside the template for coordinate idioms, follow the design system below, and run the self-review checklist before delivering.
The Cardinal Rule: CSS classes, not inline colors
The theme toggle works by switching CSS custom properties. Hardcoded fill="rgba(...)" or stroke="#22d3ee" will NOT update on theme change. Always use the class system:
<rect x="X" y="Y" width="W" height="H" rx="6" class="c-mask"/>
<rect x="X" y="Y" width="W" height="H" rx="6" class="c-backend" stroke-width="1.5"/>
<text x="CX" y="CY" class="t-primary" font-size="11" font-weight="600" text-anchor="middle">API Server</text>
<text x="CX" y="CY+16" class="t-muted" font-size="9" text-anchor="middle">FastAPI :8000</text>
Design system
Component fills c-frontend (clients/UI), c-backend (services/APIs), c-database (stores/caches), c-cloud (managed infra), c-security (auth/secrets), c-messagebus (Kafka/queues), c-external (3rd parties); text accents t-<same> plus neutrals t-primary / t-muted / t-dim. Arrows a-default, a-emphasis (hot path), a-security (dashed), a-dashed (async) - always set stroke-width and pair marker-end="url(#arrowhead[-variant])" with the matching class. Boundaries: c-security-group (dashed rose), c-region (dashed amber), c-lane (swimlane).
Typography inherits JetBrains Mono from the SVG root. Sizes: 11-12px component names, 9px sublabels, 8px annotations, 7px tiny labels.
Hard layout rules
- Two-rect pattern everywhere: opaque
c-mask rect first, styled c-<type> rect on top - semi-transparent fills otherwise let arrows bleed through.
- Arrows before components in document order (SVG paints in order; arrows must sit behind boxes).
- Vertical stacking: ≥40px gap between components; inline connectors (message buses, 20px tall) live inside the gap, never overlapping boxes.
- Boundary padding: boundary
y = inner y − 30, boundary height = inner height + 50, label baseline 18px below the boundary top.
- Legend placement: outside ALL boundary boxes, ≥20px below the lowest one; grow the viewBox if needed.
Self-review checklist (run before delivering)
grep -E 'fill="(#|rgb)|stroke="(#|rgb)' out.html inside the SVG returns nothing except the template's own defs (Cardinal Rule).- Every
c-<type> rect has an identical-geometry c-mask rect immediately before it.
- All
<line>/<path> arrows appear before all component rects in document order.
- Compute max(y + height) over all SVG elements: viewBox height must exceed it by ≥20px; same for x/width.
- Legend y is below every boundary's y + height.
- The
.toolbar, <script> blocks, and :root / [data-theme] CSS are untouched - they ARE the theme toggle and export menu.
Output
A single self-contained .html: embedded CSS with local system font fallbacks, inline SVG, and embedded theme and export JavaScript. It renders directly in any modern browser without network access. Raster exports render natively at up to 4× the viewBox, with large diagrams stepping down to stay within canvas limits. SVG downloads are dual-theme, self-contained, and follow the host's prefers-color-scheme with an optional svg[data-theme="..."] override.
