diff --git a/NEXT.md b/NEXT.md
index 1c760c2..14f5e55 100644
--- a/NEXT.md
+++ b/NEXT.md
@@ -484,10 +484,18 @@ blocks on the loader.
 
 `wait` exists for tests. A test that races the frame rate fails on a loaded
 machine, so `test/programs/agent.flan` waits for the reload instead of sleeping
-past it.
+past it. It takes **two** reloads, which is the daemon's actual loop: the first
+introduces a global the process was never built with, the second only reads it,
+and the second can only answer 1007 if it found the storage the first one
+allocated rather than a fresh zeroed copy. One reload would not have shown
+that.
 
 Two details found by running it:
 
+- **stdout is line buffered**, set in `flan_rt_init`. The C default when
+  stdout is a file or a pipe is a 4K block, so a program running with a REPL
+  attached shows nothing until it exits — and a test driving one cannot see
+  its progress at all, which is how this was found.
 - **The reply goes out before the module is queued.** The other way round, the
   game thread can install and the program can exit between the two, and the
   answer reaches the sender as a connection reset rather than as `ok`.
@@ -550,6 +558,14 @@ session over the program the process was built from, and a file of the forms
 that changed. Verified against a running sand under Xvfb — a one-form
 `game-draw` and 910 consecutive frames drew it.
 
+Two limits of that command specifically, neither of them true of sessions:
+it builds a fresh session from source on every invocation, so if the program
+file has been edited since the process launched, its idea of which names the
+host has and what its memory looks like describes a binary that is not running.
+And `Session.eval`'s `origin` defaults to `<eval>`, so an error in forms sent
+without one reports positions in a file that does not exist — the daemon has to
+pass the real buffer path, which is the same key CIDER's `eval` carries.
+
 ### What is left
 
 `C-c C-c` works end to end today; what is missing is the two hops between an
diff --git a/runtime/flan_rt.c b/runtime/flan_rt.c
index fef67da..50b61e8 100644
--- a/runtime/flan_rt.c
+++ b/runtime/flan_rt.c
@@ -26,6 +26,12 @@ static flan_slice *rt_args;   /* argv as [string], built once, never freed */
 void flan_rt_init(int32_t argc, char **argv) {
   rt_argc = (int)argc;
   rt_argv = argv;
+  /* Line buffered even when stdout is a file or a pipe, where the C default is
+   * a 4K block. A Flan program can run for minutes with a REPL attached to it,
+   * and output that only appears when it exits is output nobody can use. It is
+   * also what makes a program's progress observable to a test that is driving
+   * it. The cost is one write per line instead of per 4K. */
+  setvbuf(stdout, NULL, _IOLBF, 0);
 }
 
 void flan_argv(flan_slice *out) {
diff --git a/test/programs/agent.flan b/test/programs/agent.flan
index 73e65f8..91d1ad2 100644
--- a/test/programs/agent.flan
+++ b/test/programs/agent.flan
@@ -1,9 +1,11 @@
 ;;;; The agent, end to end: a running program takes a redefinition over a
 ;;;; socket and installs it between "frames".
 ;;;;
-;;;; [tick] is the function that gets redefined. It is called once before the
-;;;; reload and once after, and nothing else in this file changes, so the two
-;;;; numbers are the whole result.
+;;;; [tick] is the function that gets redefined. It is called once before any
+;;;; reload and once after each of two of them, and nothing else in this file
+;;;; changes, so the three numbers are the whole result. Two reloads and not
+;;;; one because that is the daemon's actual loop: a second module built from
+;;;; the same session, reading state the first one introduced.
 ;;;;
 ;;;; It waits rather than polling on a timer because a test that races the
 ;;;; frame rate is a test that fails on a loaded machine. A game loop calls
@@ -24,6 +26,8 @@
       (if (< (agent/start (at args 1)) 0)
         (do (print-line "cannot listen") 1)
         (do
+          (print-i64 (tick)) (newline)
+          (while (= (agent/wait 100) 0) 0)
           (print-i64 (tick)) (newline)
           (while (= (agent/wait 100) 0) 0)
           (print-i64 (tick)) (newline)
diff --git a/test/test_agent.ml b/test/test_agent.ml
index a85b1bf..8bcb818 100644
--- a/test/test_agent.ml
+++ b/test/test_agent.ml
@@ -77,10 +77,25 @@ let () =
       (Build.executable ~opts:dev ~csrcs:l.Load.csrcs ~lflags:l.Load.lflags
          t.Session.host ~out:exe);
 
-    (* One form, which is what C-c C-c sends. *)
-    let c = Session.eval t "(defn tick [] i64 (set ticks (+ ticks 1000)) ticks)" in
-    let so = tmp "tick.so" in
-    ignore (Build.shared ~opts:dev ~ir:c.Session.ir ~out:so ());
+    (* Two evaluations from the one session, which is the daemon's loop and
+       the thing no earlier test does. The first introduces a global the
+       process was never built with; the second only reads it, and can only
+       come back with 1007 if it found the storage the first one allocated
+       rather than a fresh zeroed copy of it. *)
+    let build_module src name =
+      let c = Session.eval t src in
+      let out = tmp name in
+      ignore (Build.shared ~opts:dev ~ir:c.Session.ir ~out ());
+      out
+    in
+    let so1 =
+      build_module
+        "(defvar acc i64) (defn tick [] i64 (set acc (+ acc 1000)) acc)"
+        "tick1.so"
+    in
+    let so2 =
+      build_module "(defn tick [] i64 (set acc (+ acc 7)) acc)" "tick2.so"
+    in
 
     let sock = tmp "sock" in
     let out = tmp "out" in
@@ -105,24 +120,38 @@ let () =
 
       (* "ok" means queued, not installed — the store happens on the other
          thread, at a time this one does not choose. *)
-      let reply = send sock so in
+      let lines () =
+        let text = In_channel.with_open_bin out In_channel.input_all in
+        List.length (String.split_on_char '\n' text) - 1
+      in
+      let reply = send sock so1 in
+      if reply <> "ok\n" then fail "agent replied %S, wanted \"ok\\n\"" reply;
+      (* Wait for the program to have consumed the first module before sending
+         the second. Both at once is a legitimate thing for the agent to do —
+         one poll installs everything queued — but then only the last one is
+         ever observed and the sequencing is not what was tested. *)
+      if not (await (fun () -> lines () >= 2)) then
+        fail "the first reload was never installed";
+      let reply = send sock so2 in
       if reply <> "ok\n" then fail "agent replied %S, wanted \"ok\\n\"" reply;
       let _, status = Unix.waitpid [] pid in
       let text = In_channel.with_open_bin out In_channel.input_all in
-      (* 1 from the original [tick], then 1001: the same call site, in a
-         program that never stopped, running a body that did not exist when it
-         started. *)
-      if status <> Unix.WEXITED 0 || text <> "1\n1001\n" then
+      (* 1 from the original [tick]; 1000 from a body that did not exist when
+         the program started, over a global that did not either; 1007 from a
+         second body that only reads it. That last number is the whole point of
+         doing this twice — a registry that handed out fresh storage per module
+         would say 7. *)
+      if status <> Unix.WEXITED 0 || text <> "1\n1000\n1007\n" then
         fail "agent reload\n        got: %S (%s)\n     wanted: %S" text
           (match status with
            | Unix.WEXITED c -> Printf.sprintf "exit %d" c
            | Unix.WSIGNALED c -> Printf.sprintf "signal %d" c
            | Unix.WSTOPPED c -> Printf.sprintf "stopped %d" c)
-          "1\n1001\n"
+          "1\n1000\n1007\n"
     end;
 
     List.iter (fun f -> try Sys.remove f with Sys_error _ -> ())
-      [ exe; so; sock; out ];
+      [ exe; so1; so2; sock; out ];
     if !failures = 0 then print_endline "agent: all tests passed"
     else begin
       Printf.printf "\n%d failure(s)\n" !failures;