Fighting broke out in western Tigray on 1 August, involving heavy artillery and drones, and drove hundreds of people across the border into Sudan. Wounded fighters were among those who reached the Sudanese town of Hamdayet, treated at a border health centre and moved on to a settlement already housing people displaced by the war of 2020 to 2022. The clashes ended the same day. A Tigray People’s Liberation Front official told AFP the two sides had de-escalated. Federal forces have since reinforced western Tigray and set up checkpoints restricting movement outside Humera, and residents of Alamata, further south, describe a military build-up that has them expecting the war to resume.
A one-day battle that both sides walked back is easy to file as a scare. That filing misreads which instrument is actually dismantling the peace. The Pretoria agreement, brokered by the African Union in November 2022, committed Ethiopia to restoring constitutional order and returning occupied territories to Tigray. It did not specify how, or by when, and it left the status of the contested districts to a later process that was never convened. In February 2026, the Raya districts including Alamata were removed from Tigrayan oversight and reassigned as electoral constituencies ahead of national polls. No offensive was required. A constituency list did what three years of unresolved negotiation had not, and it did so with the authority of routine electoral administration.
This is the mechanism worth naming. Territory in Ethiopia is not held by whoever occupies it on a given morning. It is held by whichever administration registers the voters, appoints the officials, collects the revenue and draws the boundary on the map the federal government uses. Those functions are exercised through paperwork, and paperwork is not covered by a ceasefire. A peace agreement that stops the shooting while leaving the administrative apparatus in the hands of one party has not frozen the dispute. It has handed one side a slower and more durable instrument than the one it put down.
A constituency list did what three years of unresolved negotiation had not, and it did so with the authority of routine electoral administration.
The districts in question are not incidental. Alamata is the hub of the Raya valley, which straddles the Tigray and Amhara regions and sits on the strategic highway linking Tigray to central Ethiopia. Amhara ethno-nationalists claim the area. Hundreds of thousands of Tigrayans were driven out of these districts during the war. Reassigning them for electoral purposes therefore does more than adjust a boundary. It fixes in law a demographic outcome produced by force, and it does so at the moment when the displaced have not returned and cannot vote on it. Analysts have described the February decision as war preparation conducted by administrative means, and the description is hard to argue with once the sequence is laid out.
Tigray’s own politics have run in the same direction. The TPLF reinstated the pre-war regional government in May 2026, a step that voided the agreement from its side and removed the dispute-resolution channel the text had established. Its forces took control of Alamata at the start of this year after federal troops withdrew. Both parties have now acted outside the framework they signed, and each can point at the other’s breach as justification for its own. What has disappeared in the process is not goodwill. It is the last venue in which either party could have raised a grievance and expected a ruling.
That absence is the African Union’s problem as much as Ethiopia’s. Pretoria is the continental body’s most cited peacemaking success, invoked whenever the case is made for African solutions to African conflicts. The agreement did stop a war that was killing people at industrial scale, and that achievement is real. But it was built on deferral. The hardest question, who governs the contested territory and on what basis, was postponed rather than answered, and the party holding the administrative machinery has spent the interval answering it unilaterally. A mediation architecture that treats the cessation of hostilities as the deliverable and the territorial settlement as follow-up work will keep producing agreements with this failure mode built in.
The signals now point one way. Federal reinforcements are moving into western Tigray. Checkpoints are constraining movement around Humera. Civilians in Alamata are reading the build-up as a prelude and saying so. Refugees are already crossing into a Sudan that is fighting its own war and can absorb nobody. Political divisions inside the TPLF add a further variable, because a fragmented party is harder to bind to any settlement its leadership might accept.
If the shooting resumes at scale, the account will date the collapse to a firefight and treat the paperwork as background. That will be the wrong chronology. The Pretoria agreement was not broken on 1 August in western Tigray. It was broken in February in an electoral administrative decision that required no soldiers, drew no international response, and settled the question the treaty had refused to settle. The fighting, when it comes, will be the confirmation rather than the cause.
