diff --git a/code/integration-tests/runtime-tests/README.md b/code/integration-tests/runtime-tests/README.md
new file mode 100644
index 00000000000..9f13eb34b00
--- /dev/null
+++ b/code/integration-tests/runtime-tests/README.md
@@ -0,0 +1,7 @@
+# Priorities
+
+1. No false negatives. Such tests to be deleted (see rfc about brittle and flaky tests)
+2. Time to merge into main. Should be fast.
+3. Super low maintainance. 20% of tests for critical paths cover 80% of failures. Easy to find why in the internet why. Use same tools as ops and devs.
+4. No bugs into main merged.
+5. Readability of tests output. Easy to reproduce tests locally and doing smoke tests in repeatable way so can just eyeball what wrong.
diff --git a/code/integration-tests/runtime-tests/runtime-tests.nix b/code/integration-tests/runtime-tests/runtime-tests.nix
index 396485897d8..3a58f7044b2 100644
--- a/code/integration-tests/runtime-tests/runtime-tests.nix
+++ b/code/integration-tests/runtime-tests/runtime-tests.nix
@@ -17,7 +17,10 @@
           } 2>&1 & ) | tee devnet-xc.log &
 
           process-compose-stop() {
-            for i in $(process-compose process list); do process-compose process stop "$i"; done
+            for i in $(process-compose process list)
+            do
+              process-compose process stop "$i"
+            done
           }
 
           TRIES=0