diff --git a/.graphite.yml b/.graphite.yml
index 33b1da0..63bd4a6 100644
--- a/.graphite.yml
+++ b/.graphite.yml
@@ -1,49 +1,47 @@
 projects:
-    - slug: ma'at
-      kind: project
-      name: Ma'at
-      summary: >
-          Ma'at is a distributed online judge that runs untrusted C++ and Python code in locked-down Docker sandboxes, with horizontally scalable workers, live verdicts over WebSockets, and contest leaderboards.
+  - slug: maat
+    kind: project
+    name: Ma'at
+    summary: >
+      Ma'at is a distributed online judge that runs untrusted C++ and Python code in locked-down Docker sandboxes, with horizontally scalable workers, live verdicts over WebSockets, and contest leaderboards.
+    tags: [javascript, mongodb, redis, react, nodejs, docker]
+    owner: omar-mohamed24
+    authors: [Omar-Mohamed24]
+    status: wip
+    listed: false
+    pinned: false
+    showcase:
+      problem: >
+        Practicing for programming contests usually means relying on big public judges, and
+        building your own for a class or a contest is hard because it means safely running
+        untrusted code from strangers. There was no simple self-hostable judge that isolates
+        submissions, handles many at once and shows results live.
+      approach: >
+        A distributed online judge named after the Egyptian goddess of truth and justice. A
+        Node.js and Express API stores submissions in MongoDB and queues them in Redis with BullMQ.
+        Stateless workers pull jobs and compile and run C++ and Python code in
+        throwaway Docker sandboxes with no network, a read-only filesystem, and capped CPU,
+        memory and processes, then classify the verdict against hidden test cases. Results
+        reach a React frontend over WebSockets, and Redis sorted sets power live contest
+        leaderboards. Docker Compose and GitHub Actions CI/CD handle deployment.
+      outcome: >
+        Users submit code and see a verdict (Accepted, Wrong Answer, Time Limit, Memory Limit,
+        Runtime or Compile Error) within seconds, with no page refresh. Capacity scales by
+        adding workers, and a k6 load test measured [X] submissions per minute with [N]
+        workers. It hosted a contest for [N] real participants.
+      contributions:
+        - github: Omar-Mohamed24
+          summary: >
+            Built the whole platform: the sandboxed judge workers, the queue-based API, the
+            React frontend with live verdicts, and the CI/CD and Docker Compose deployment.
+    deploy:
+      kind: none
 
-      tags: [javascript, mongodb, redis, react, nodejs, docker]
-      owner: omar-mohamed24
-      authors: [Omar-Mohamed24]
-      status: wip
-      listed: false
-      pinned: false
-      showcase:
-          problem: >
-              Practicing for programming contests usually means relying on big public judges, and
-              building your own for a class or a contest is hard because it means safely running
-              untrusted code from strangers. There was no simple self-hostable judge that isolates
-              submissions, handles many at once and shows results live.
-          approach: >
-              A distributed online judge named after the Egyptian goddess of truth and justice. A
-              Node.js and Express API stores submissions in MongoDB and queues them in Redis with BullMQ.
-              Stateless workers pull jobs and compile and run C++ and Python code in
-              throwaway Docker sandboxes with no network, a read-only filesystem, and capped CPU,
-              memory and processes, then classify the verdict against hidden test cases. Results
-              reach a React frontend over WebSockets, and Redis sorted sets power live contest
-              leaderboards. Docker Compose and GitHub Actions CI/CD handle deployment.
-          outcome: >
-              Users submit code and see a verdict (Accepted, Wrong Answer, Time Limit, Memory Limit,
-              Runtime or Compile Error) within seconds, with no page refresh. Capacity scales by
-              adding workers, and a k6 load test measured [X] submissions per minute with [N]
-              workers. It hosted a contest for [N] real participants.
-          contributions:
-              - github: Omar-Mohamed24
-                  summary: >
-                      Built the whole platform: the sandboxed judge workers, the queue-based API, the
-                      React frontend with live verdicts, and the CI/CD and Docker Compose deployment.
-
-      deploy:
-          kind: none
-
-      # deploy:
-      #   kind: compose
-      #   host: maat.project-graphite.com
-      #   source:
-      #     file: compose.production.yaml
-      #     service: frontend
-      #     port: 80
-      #   secrets: []
+    # deploy:
+    #   kind: compose
+    #   host: maat.project-graphite.com
+    #   source:
+    #     file: compose.production.yaml
+    #     service: frontend
+    #     port: 80
+    #   secrets: []