Sethu Nguna.
Independent project · Strategy game & RPG · 2025–2026

African Mandate

Public demo · in development

You are the African Union’s Special Envoy to the Sahel. Over 20 turns spanning 48 months, you try to stabilise a region where insurgency, coups, climate stress and outside influence compound one another, and where partial success is the realistic win.

← Explore all projects
Audience
Strategy players, and policy and governance professionals
Sahel Arena
Five countries, 30 interactive zones, three actions a turn, five regional metrics and five endings
The RPG
An evidence-governed rebuild: real conflict and finance data, simulated actors labelled as simulated, no overall score
Built with
React, TypeScript, Vite and Leaflet; browser saves, with experimental cloud saves on Supabase; Playwright release gates
Sahel Arena situation room: turn progression, resources and regional metrics on the left, a zone map of the Sahel in the centre, and an intelligence feed on the right
Sahel Arena’s situation room in the public desktop demo, which runs on synthetic data.
The design problem

The challenge

The Sahel’s crisis doesn’t fit a lecture. Security, governance, climate and legitimacy pull against one another, and the instruments available to an envoy are limited. Policy audiences need to feel those trade-offs, not just read about them.

The logic

Put the player in an African Union role rather than an outside power’s, with scarce budget, political capital and personnel, and make every action trade one outcome against another. As the game puts it: you cannot airstrike a drought.

What players produce

A 20-turn campaign and one of five endings, from stabilisation to the collapse of the mandate.

My role

I designed, wrote and developed both builds: the systems design, the narrative and the code.

Two builds

Sahel Arena

The v0.1 public web release for desktop browsers. Players start as a guest with saves kept in the browser and play through narrative acts against AI-driven insurgents. Unsupported setups are stopped before a campaign starts, rather than failing halfway through.

African Mandate RPG

A rebuild with explicit evidence boundaries. An intelligence-informed model separates observed historical evidence, latent scenario conditions and what the player actually knows. Zones come from Natural Earth, conflict events and development-finance records are cited and hashed, and missing observations stay unknown instead of becoming zero.

How a turn plays

Read, engage, act, live with it.

Twenty turns, each one a month or more of the mandate.

01

Read the feed

Intelligence arrives with urgency and confidence signals: threat assessments, insurgent movements, smuggling surges.

02

Engage the actors

The African Union Commission, ECOWAS, transitional governments, the donor community and civil society, each with its own readiness to talk.

03

Take three actions

Security patrols, diplomacy, governance reform, aid, intelligence operations or community mediation, paid for in budget, political capital and personnel.

04

Watch the metrics

Stability, insurgency, civilian support, global legitimacy and regional synergy move, rarely all in the same direction.

05

End the turn

Events land: insurgent surges, corruption scandals, donor pullouts. Plans made last turn may no longer hold.

06

Reach an ending

After 20 turns the campaign resolves into one of five endings.

Design decisions

Every mechanic has a job.

Both builds are designed so that players learn the limits of what an envoy can do.

DecisionWhat it trainsIn the game
An African Union envoy, not an outside power Seeing the crisis from the institution that holds the mandate. The player is the AU Special Envoy for the Sahel.
Three actions a turn Ruthless prioritisation. Budget, political capital and personnel are never enough for everything.
Five metrics that pull apart Trade-offs instead of a single objective. Stability, insurgency, civilian support, global legitimacy and regional synergy.
Confidence on every report Acting on imperfect intelligence. Each feed item carries urgency and confidence signals.
Evidence, conditions and knowledge kept apart Knowing what the model knows and what it assumes. In the RPG, observed data, latent scenario conditions and player knowledge are modelled separately.
Simulated actors, labelled Not mistaking a game for a forecast. In the RPG, fictional procedures and actor positions carry a visible “simulated” label, with no real person or asserted private intent.
Trajectory, not a score Judging a mandate by how it went, not by one number. The RPG weighs the end state at 40% and the trajectory at 60%, penalises prolonged crises and countervailing harm, and shows no overall score or leaderboard.
Evidence

What’s documented

Sahel Arena v0.1 is publicly playable on desktop browsers, and each release has to pass automated end-to-end tests in Chrome and Edge. The RPG’s build runs through a logged acceptance process, with owner review at every stage.

Limits

What the evidence can’t show

Neither build is an operational forecasting system. Sahel Arena’s public demo runs on synthetic data. The RPG isn’t playable yet: its production geography and opening campaign values are still awaiting sign-off. No player testing has been reported.

Supporting artifacts

Explore the work.

Start a conversation

Build the capability.
Make it work in practice.

Learning strategy, institutional systems and decision practice.

sethnguna@gmail.com ↗