diff --git a/csaxs_bec/bec_ipython_client/plugins/LamNI/thingstotest.txt b/csaxs_bec/bec_ipython_client/plugins/LamNI/thingstotest.txt
new file mode 100644
index 0000000..9c53203
--- /dev/null
+++ b/csaxs_bec/bec_ipython_client/plugins/LamNI/thingstotest.txt
@@ -0,0 +1,37 @@
+1. Things to test
+Progress persistence
+
+Start tomo_scan(), let a few projections run, restart the BEC client mid-scan, confirm client.get_global_var("tomo_progress") still shows heartbeat/tomo_start_time from before the restart
+After ~10 projections, confirm CLI output shows a real ETA (not "N/A") — the threshold is projection > 9
+Confirm the GUI progress display (lamnigui_show_progress()) shows a real start time / ETA, not "N/A"
+Force a gap between projections (pause, or simulate an interlock) and confirm the "Detected a X gap... excluding Y from ETA" message appears and accumulated_idle_time grows
+Call lamni.tomo_progress_reset(), confirm it clears and prints confirmation
+
+Correction persistence
+
+lamni.read_additional_correction(<file>), restart the client, confirm lamni.corr_pos_x/corr_pos_y/corr_angle (+ _2) still hold the loaded values
+lamni.reset_correction() / reset_correction_2() — confirm they clear to [] and that clears persist across a restart too
+Confirm lamni.compute_additional_correction(angle) still returns sensible values after a restart
+
+X-ray eye alignment (full run)
+
+Run lamni.xrayeye_alignment_start() end-to-end
+Crosshair appears at the FZP centre after step 0, stays put through all angle steps, disappears at completion
+Ctrl-C mid-alignment — confirm the crosshair still disappears (abort path)
+Confirm dev.fsh physically opens/closes correctly (not a leftover omnyfsh reference anywhere)
+Open the resulting xrayeye_alignmentvalues_<timestamp>.h5: check alignment_values (9,2), alignment_images has real frames, roi_pixel_data has one row per submit, alignment_fit is (3,8)
+Confirm the old plain-text file is still written unchanged alongside it
+keep_shutter_open=True path still prompts and closes correctly
+Watch specifically for whether click positions land where expected (the un-verified /2 pixel-calibration question)
+
+Burst mode
+
+Set lamni.frames_per_trigger = N (N>1), run one projection, confirm the detector actually acquires N frames per point
+Confirm tomo_parameters() shows and lets you edit it
+Confirm 0, negative, non-int, or bool raise ValueError
+Confirm leaving it unset still behaves as 1 (no behavior change for existing scans)
+
+Regression
+
+extra_tomo.py: exercise MagLamNI.rotate_slowly() — confirm no more AttributeError on self.align
+One full short tomo_scan() end-to-end, to confirm none of the above broke the main scan path
\ No newline at end of file