diff --git a/cli/src/commands/exec/mod.zig b/cli/src/commands/exec/mod.zig
index efa873a..5602df9 100644
--- a/cli/src/commands/exec/mod.zig
+++ b/cli/src/commands/exec/mod.zig
@@ -50,7 +50,7 @@ pub fn run(allocator: std.mem.Allocator, args: []const []const u8) !void {
         const arg = pre[i];
 
         if (std.mem.eql(u8, arg, "--help") or std.mem.eql(u8, arg, "-h")) {
-            try printUsage();
+            std.debug.print("Exec module - used internally by 'ml run'\n", .{});
             return;
         } else if (std.mem.eql(u8, arg, "--json")) {
             flags.json = true;
@@ -95,7 +95,7 @@ pub fn run(allocator: std.mem.Allocator, args: []const []const u8) !void {
     }
 
     if (job_name == null) {
-        try printUsage();
+        std.log.err("No job name specified", .{});
         return error.InvalidArgs;
     }
 
@@ -159,36 +159,3 @@ pub fn run(allocator: std.mem.Allocator, args: []const []const u8) !void {
         },
     }
 }
-
-fn printUsage() !void {
-    std.debug.print(
-        \\n
-        \\ml exec <job_name> [options] [-- <args>]
-        \\
-        \\Unified execution - works locally or remotely with transparent fallback.
-        \\
-        \\Options:
-        \\  --priority <1-10>       Job priority (default: 5)
-        \\  --cpu <n>               CPU cores requested (default: 1)
-        \\  --memory <n>            Memory GB requested (default: 4)
-        \\  --gpu <n>               GPU devices requested (default: 0)
-        \\  --gpu-memory <spec>     GPU memory spec
-        \\
-        \\Execution mode is controlled by execution_mode setting in config.
-        \\Use 'ml init --mode=local|remote|auto' to change.
-        \\
-        \\Research context:
-        \\  --hypothesis <text>     What you're testing
-        \\  --context <text>        Background information
-        \\  --intent <text>         What you're trying to do
-        \\  --expected-outcome <t>  What you expect to happen
-        \\  --tags <csv>            Comma-separated tags
-        \\
-        \\Examples:
-        \\  ml exec train.py
-        \\  ml exec train.py -- --lr 0.001 --epochs 10
-        \\  ml exec train.py --priority 8 --gpu 1
-        \\  ml exec train.py --hypothesis "LR scaling helps"
-        \\
-    , .{});
-}