An executive dashboard for H.E. and senior leadership to review the AI agents designed across government departments. The interface follows the UAE Government Design System — restrained colour, strong spacing, clear hierarchy, accessible contrast, rounded cards and soft elevation — so it reads as an official UAE government internal transformation tool rather than a generic SaaS dashboard.
This is a static site, so GitHub Pages serves it directly. One-time setup
(needs repo admin): Settings → Pages → Source: "Deploy from a branch" →
Branch: claude/nice-allen-uqtnf6 → folder / (root) → Save. After ~1 minute
the dashboard is live at:
https://txlabtesting.github.io/agentified/
It re-deploys automatically on every push. (A .nojekyll file is included so
all assets are served as-is.)
No build step, no backend.
# from the project root
python3 -m http.server 8099
# then open http://localhost:8099Or simply open index.html in a browser.
- How many agents do we have overall, and across how many departments?
- Which departments have the most agents?
- Which agents are complex, and what is the average complexity?
- Which agents need review, update or refinement?
- Which departments are ready?
- What does each agent do, and which sub-agents sit under it?
- What should leadership update or approve next?
| Page | What it shows |
|---|---|
| Overview | Six KPI cards, agents per department bar chart, complexity donut, status distribution, and a sortable department summary table. |
| Departments | Department cards with agent/sub-agent counts, complexity distribution and a readiness score. |
| Department detail | Department description, headline stats, owner/focal point, and every main agent as a rich card (complexity, impact, feasibility, status, priority, sub-agent count). |
| Agent detail drawer | Right-side drawer: purpose, responsibilities, process, inputs, systems, outputs, autonomy, risks/dependencies, recommended next action, and the full sub-agent table. |
| Agents | Every agent across departments in one filterable, complexity-sorted table. |
| Sub-Agents | Every sub-agent with parent, task type, complexity, dependencies and status. |
| Pending Review | Agents flagged Needs Review, grouped by department, highest complexity first. |
| Agent Assistant | A chat tester to query the agent inventory in plain language. |
| Settings | Dataset overview and data actions (export, reset edits). |
A conversational assistant (under Tools → Agent Assistant) to quickly test and explore the agent list. It is grounded in the live dataset — including any in-session edits — so it always reflects exactly what is loaded. Ask things like:
- How many agents do we have? · Which department has the most agents?
- Which agents need review? · Show high-complexity agents
- Tell me about the Payroll Validation agent (opens a profile + drawer link)
- What sub-agents are under Onboarding Orchestration?
- Which departments are ready? · List the value-add agents
It runs entirely client-side — no API key — so it works on the published link.
API-ready: the engine lives in assets/js/assistant.js
behind a clean seam. To swap in a real LLM later, stand up a small serverless
function that calls the Claude API (key stays server-side), then set:
Assistant.config.mode = "api";
Assistant.config.endpoint = "/api/chat"; // returns { answer: "<html|text>" }The grounded engine remains as an automatic offline fallback.
- Search across agents, sub-agents and departments (live).
- Filter by type, complexity, status and priority (combinable).
- Drill down from KPI charts and the department table into a department, then into an agent.
- Edit / Update Agent — an in-session modal to update name, department, complexity, impact, feasibility, status, priority, sub-agents, notes and recommended action.
- Export the full agent inventory to CSV; Share copies the dashboard link.
- Fully responsive (desktop → tablet → mobile) with a collapsible sidebar.
- Every department is authored from its live Agentic Blueprint (01–10): Human Resources, Procurement/Travel/Vendor, Finance & Accounting, Knowledge & Content, Legal, Events & Communication, Cyber Security, IT Operations, Protocol and Admin Services — each with its real complexity / impact / feasibility / phase ratings, purpose, responsibilities, systems, autonomy and recommended actions, plus derived sub-agents.
- Totals (143 main agents · 427 sub-agents · 10 departments) are computed at runtime, so refreshing the data later requires no UI changes.
All data lives in assets/js/data.js; the UI layer is in
assets/js/app.js and styling tokens in
assets/css/styles.css.
Prototype: edits are held in memory for the session only and are not persisted.