// Render interactive React artifacts for EA interview sessions and brainstorming. Use the interview app when conducting structured Q&A for an artifact or phase. Use the brainstorm pad when capturing freeform engagement thoughts.
This skill should be used when the user asks to "create an ArchiMate diagram", "model this in ArchiMate", "what ArchiMate element is this", "show the ArchiMate layers", "represent this architecture using ArchiMate", or when producing architecture diagrams using ArchiMate 3.x notation. Provides element types, layers, relationships, and diagram guidance including Mermaid, Graphviz, and Draw.io representations.
This skill should be used when the user asks for an "architect lens", "practitioner review", "seasoned architect review", "engagement health from a practitioner perspective", or invokes /ea-lens. Provides an opinionated engagement-level review from the perspective of a senior EA practitioner focused on what actually matters — not TOGAF compliance theatre.
Manages the shared TOGAF Architecture Repository — Standards Information Base (SIB/STD-NNN), Vendor Landscape Register (VDR-NNN), Technology Horizon Register (THR-NNN), and enterprise-level governance artefacts. Supports multi-engagement, multi-project sharing via the EA-Workspace/ sibling-folder layout.
This skill should be used when the user asks to "create an artifact", "generate the architecture vision", "start a new artifact from a template", "what template should I use", "populate this artifact", or when any TOGAF artifact needs to be created or populated. Provides template selection, placeholder conventions, and guidance text marking standards for all EA artifacts.
This skill should be used when the user asks to "manage architecture constraints", "add a constraint", "view constraints", "trace a constraint to an artifact", "update the constraints register", or "assess constraint impact". Handles the full constraints lifecycle from capture through traceability and sync with a shared constraints repository.
This skill should be used when the user asks to "upload a document", "import a diagram", "read this Word file", "use this PDF as input", "parse the completed interview", "import my answers from Word", "ingest an existing architecture document", or when processing uploaded files to extract EA-relevant content for use in artifacts or interviews.
| name | ea-interview-ui |
| description | Render interactive React artifacts for EA interview sessions and brainstorming. Use the interview app when conducting structured Q&A for an artifact or phase. Use the brainstorm pad when capturing freeform engagement thoughts. |
| version | 0.9.55 |
Provides two interactive UI apps for EA interview sessions and brainstorming.
This skill is only invoked when the user selects Web mode. The default interview mode is Text (chat Q&A). Only load this skill when mode = "web" has been explicitly chosen.
Runtime detection — read uiMode from .claude/ea-assistant.local.md first, then choose the delivery mode:
uiMode setting | Delivery mode |
|---|---|
artifact | React artifact — present JSX inline; use in Claude Code Desktop, Cowork Desktop, or claude.ai/code |
html | HTML file — write to disk and open in browser; works in all environments including CLI terminal |
| not set | Default to HTML file |
Check for MCP mode before either: if the preview_artifact tool is listed in the available tools, use it regardless of uiMode — pass the completed HTML string and the artifact/session title. The server handles opening the browser.
Both apps are stateless — all state lives in React useState. A page refresh resets the form, so complete the session in one sitting before copying the output.
When to use: When the user selects Web mode for an interview session. Not used for Text or Display modes.
references/interview-app-template.jsx.INTERVIEW_DATA constant (lines beginning const INTERVIEW_DATA = {) with the actual session data:const INTERVIEW_DATA = {
artifactName: "<artifact or phase name>",
engagementName: "<engagement name from engagement.json>",
mode: "artifact", // or "phase"
questions: [
{
id: "q1",
text: "<question text extracted from {{placeholder}} or phase question bank>",
context: "<one-sentence explanation of why this matters>",
defaultAnswer: "<single best-practice default, or null>",
existingAnswer: "<value from previous session, or null>",
brainstormNote: "<relevant thought from brainstorm-notes.md, or null>",
suggestions: [
{ label: "<short chip label>", value: "<full answer text the user can edit>" },
// 2–4 entries covering common good-practice answers for this question in context
// null if no common answers are relevant
],
options: ["<option 1>", "<option 2>"], // or null — for checklist questions
allowMultiple: true, // false for single-select (radio); omit if options is null
},
// ... one entry per question
],
};
Set options when the question bank has enumerated choices (e.g. constraint types, data migration approaches, cut-over approaches). Set allowMultiple: false for single-select questions (radio buttons).
Set suggestions when common good-practice answers exist for the question given the engagement context. Suggestions are rendered as clickable chips labelled with label; clicking one loads the full value text into the textarea so the user can edit it before submitting. Use 2–4 suggestions per question. Suppress suggestions when existingAnswer is set (the user already has a prior answer to work from). Do not set both suggestions and options on the same question — use options for constrained enumerated choices and suggestions for open-ended questions with common starting points.
Set voiceEnabled: true when the user selects Voice mode. This adds a 🎤 microphone button to every question card:
window.SpeechRecognition (or window.webkitSpeechRecognition); button shows 🔴 Recording…references/interview-app-shell.html.%%ARTIFACT_NAME%% with the artifact or phase name.%%INTERVIEW_DATA_JSON%% with the JSON-serialised INTERVIEW_DATA object (same shape as above).%%COMPONENT_BODY%% with the full body of references/interview-app-template.jsx — everything after the import line and INTERVIEW_DATA constant (i.e. from const ANSWER_STATE to the end of the file), excluding the final export default keyword (change export default function InterviewApp → function InterviewApp).EA-projects/{slug}/ui/interview-{YYYY-MM-DD}.html.cmd.exe /c start "" "$(wslpath -w EA-projects/{slug}/ui/interview-{YYYY-MM-DD}.html)"xdg-open / openApp behaviour (for reference — do not re-implement in instructions):
Processing the paste-back:
When the user pastes the results block, parse each line:
Q{N} [Answered]: ... + Answer: ... → write the answer to the artifact fieldQ{N} [Default accepted]: ... + Answer: ... → write the answer, append ✓ Default acceptedQ{N} [Skipped]: ... → write ⚠️ Not answeredQ{N} [N/A]: ... → write ➖ Not applicableThen save interview notes and update engagement.json as per the normal interview completion flow.
When to use: At the start of every ea-brainstorm command session, in place of the iterative chat loop.
The brainstorm pad is parameterised via a BRAINSTORM_DATA constant injected at render time (same pattern as INTERVIEW_DATA in App 1). When no phase is provided, all fields are null and the app shows generic hints.
const BRAINSTORM_DATA = {
phase: "<phase label, e.g. 'Phase D — Technology Architecture'>", // null = no badge
subtitle: "<one-sentence focus prompt>", // null = generic subtitle
categories: [ // null = use all generic defaults
// --- Override an existing default category ---
{ id: "concerns", hint: "<phase-specific hint text>",
suggestions: ["<common concern for this phase>", "<another common concern>"] }, // or omit/null
{ id: "goals", hint: "<phase-specific hint text>",
suggestions: ["<common goal>", "<another common goal>"] },
{ id: "constraints", hint: "<phase-specific hint text>",
suggestions: null },
{ id: "opportunities", hint: "<phase-specific hint text>",
suggestions: null },
{ id: "assumptions", hint: "<phase-specific hint text>",
suggestions: null },
{ id: "other", hint: "<phase-specific hint text>",
suggestions: null },
// --- Add a new custom category (creates an extra card) ---
{ id: "value-streams", label: "Value Streams", emoji: "🌊",
hint: "End-to-end chains that deliver stakeholder value",
suggestions: ["Order-to-cash value stream fragmented across 3 systems"] },
{ id: "platforms", label: "Platforms", emoji: "🏗️",
hint: "Runtime platforms and execution environments",
suggestions: null },
],
questions: [ // null if no artifact/phase context — upcoming interview questions shown as prompts
{
id: "q1",
text: "<question that will be asked in the upcoming interview>",
category: "concerns | goals | constraints | opportunities | assumptions | other | value-streams | use-cases | processes | conceptual-model | logical-model | platforms | languages | infrastructure | network",
},
// ... up to 8 questions
],
prefilled: [ // null if no pre-existing answers — document/artifact answers pre-populated in pad
{
questionRef: "<placeholder key or question id>",
questionText: "<full question text>",
answer: "<pre-existing answer from document or artifact>",
source: "<'artifact' or upload filename>",
category: "concerns | goals | constraints | opportunities | assumptions | other | value-streams | use-cases | processes | conceptual-model | logical-model | platforms | languages | infrastructure | network",
},
// ...
],
};
Only hint and suggestions are overridden for default ids — label and emoji are always taken from DEFAULT_CATEGORIES in brainstorm-app.jsx. For new ids (not in the default set), the app creates an additional category card using the label, emoji, hint, and suggestions provided. If label is omitted, the id is converted to title-case. If emoji is omitted, it defaults to 📋.
App behaviour for questions: Within each category card, show a collapsible "📋 Upcoming questions" section (collapsed by default). List each question whose category matches the card. This gives the user context for what topics to brainstorm before the interview.
App behaviour for prefilled: Within each category card, pre-populate a thought entry for each prefilled item whose category matches. Tag each pre-filled entry with a 📄 From: {source} badge. The user can:
📄 tag)📄 tag is retained so Claude can detect modifications at paste-back)references/brainstorm-app.jsx.BRAINSTORM_DATA constant (lines beginning const BRAINSTORM_DATA = {) with the actual data provided by ea-brainstorm.md. When no phase is active, set all fields to null.references/brainstorm-app-shell.html.%%BRAINSTORM_DATA_JSON%% with the JSON-serialised BRAINSTORM_DATA object (same shape as above).%%COMPONENT_BODY%% with the full body of references/brainstorm-app.jsx — everything after the import line and BRAINSTORM_DATA constant (i.e. from const DEFAULT_CATEGORIES to end of file), changing export default function BrainstormPad → function BrainstormPad.EA-projects/{slug}/ui/brainstorm-{YYYY-MM-DD}.html.App behaviour (for reference):
id in BRAINSTORM_DATA.categories that is not in the base set (e.g., Value Streams, Platforms)BRAINSTORM_DATA.categories is setBRAINSTORM_DATA.phase is setPhase: line when scoped); user copies and pastes it into chatProcessing the paste-back:
When the user pastes the BRAINSTORM NOTES block, parse the categories and thoughts, then:
[📄 From: {source}] tag in the thought textea-brainstorm.md step 5b for conflict detection before savingEA-projects/{slug}/brainstorm/brainstorm-notes.mdea-brainstorm.md/ea-interview."Both apps:
useState, useEffect, useRef) — no external CSS or icon librariesuseState; a page refresh resets the formuiMode: artifact): no CDN or build step required — Claude Code Desktop, Cowork Desktop, and claude.ai/code bundle React. Not available in Claude Code CLI terminal (no artifact renderer).uiMode: html, default): requires an internet connection for CDN scripts (React 18 + Babel Standalone from unpkg); total ~1 MB. Works in all environments including CLI terminal.<script type="text/babel"> so the JSX is transformed at runtime — no pre-compilation neededuiMode unset or set to html — UIs will open as a standalone HTML file in your browseruiMode: artifact to render UIs inline as React artifacts