From 91ba03fd8476fe10308bd4f4a4a9c67cde39e50e Mon Sep 17 00:00:00 2001
From: Jammy2211 <JNightingale2211@gmail.com>
Date: Mon, 7 Sep 2026 17:04:08 -0400
Subject: [PATCH] cortex: retire inference_programme (drop 5 tasks, stash RAL
 outputs)

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01HUNchbL2Ugr3fPfXquLALR
---
 dashboard.html                                | 30 +++-----
 dashboard.md                                  | 72 ++++---------------
 projects.yaml                                 |  4 +-
 rulings/2026/09/R-20260907-02.md              | 21 ++++++
 rulings/2026/09/R-20260907-03.md              | 22 ++++++
 rulings/2026/09/R-20260907-04.md              | 21 ++++++
 rulings/2026/09/R-20260907-05.md              | 17 +++++
 rulings/2026/09/R-20260907-06.md              | 19 +++++
 .../cluster_extended_source_inference.md      |  6 +-
 .../legacy_point_output_sweep.md              |  6 +-
 .../likelihood_term_levers.md                 |  6 +-
 .../nautilus_mass_pix_baseline.md             |  6 +-
 tasks/inference_programme/prodigy_mass_pix.md |  6 +-
 13 files changed, 139 insertions(+), 97 deletions(-)
 create mode 100644 rulings/2026/09/R-20260907-02.md
 create mode 100644 rulings/2026/09/R-20260907-03.md
 create mode 100644 rulings/2026/09/R-20260907-04.md
 create mode 100644 rulings/2026/09/R-20260907-05.md
 create mode 100644 rulings/2026/09/R-20260907-06.md

diff --git a/dashboard.html b/dashboard.html
index e80be28..df37e25 100644
--- a/dashboard.html
+++ b/dashboard.html
@@ -227,7 +227,7 @@
 </head>
 <body>
 <header class="hero"><span class="orb"><svg class="mark" viewBox="0 0 48 48" fill="none" stroke="currentColor" stroke-width="1.3" stroke-linecap="round" stroke-linejoin="round" aria-hidden="true"><circle cx="24" cy="24" r="20.4"/><path d="M11.8,18.6 C14.4,14.8 17.8,14.4 20.1,17.1 C22.4,19.8 25.8,20.2 28.5,17.9 C31.2,15.6 34.4,16.2 36.1,19.1"/><path d="M11.6,24.6 C14.2,20.8 17.6,20.4 19.9,23.1 C22.2,25.8 25.6,26.2 28.3,23.9 C31.0,21.6 34.2,22.2 35.9,25.1"/><path d="M12.8,30.6 C15.3,27.2 18.4,26.8 20.5,29.4 C22.6,32.0 25.7,32.4 28.2,30.2"/><circle cx="32.6" cy="33.0" r="3.8"/><g fill="currentColor" stroke="none"><circle cx="32.6" cy="33.0" r="1.5"/><circle cx="11.7" cy="18.6" r="1.2"/><circle cx="11.5" cy="24.6" r="1.2"/></g></svg></span><h1>PyAuto<b>Cortex</b><span class="kind">Dashboard</span></h1><div class="rule"></div><p class="tag">Question. Run. Rule.</p></header><p class="lede">Every science task the Cortex is holding, one line each under its project. Tap a task's 📋 and its next command is on your clipboard. The verdict is always yours — this page hands you the command, never the ruling.</p>
-<ul class="stats"><li><b>0</b><span>Awaiting ruling</span></li><li><b>28</b><span>Recent rulings</span></li></ul>
+<ul class="stats"><li><b>0</b><span>Awaiting ruling</span></li><li><b>33</b><span>Recent rulings</span></li></ul>
 <div class="fresh"><p class="muted">Last updated 2026-09-07.</p></div>
 <p class="muted mdsrc"><a href="https://github.com/PyAutoLabs/PyAutoCortex/blob/main/dashboard.md">markdown version</a> · <a href="https://github.com/PyAutoLabs/PyAutoCortex/blob/main/README.md">GitHub Page</a></p>
 <div class="task"><button class="copy" data-cmd="/cortex — check in on every active science project since the last check-in (2026-09-07T20:42Z): pull each project through its sync CLI, score every live run against its pre-registered witness, move what came back, re-render and push the board, then read me the by-project summary with the prompt each task needs next." aria-label="Copy the Claude command">📋</button><p>check in since last time</p></div>
@@ -237,7 +237,6 @@
 <table class="map">
 <tr><th>Project</th><th>Next task</th><th>Awaiting</th><th>Running</th><th>Last ruling</th></tr>
 <tr><td><b>euclid_dr1_prelim</b></td><td>Ten real DR1 lenses fitted from the pipeline repo alone</td><td>0</td><td>1</td><td>-</td></tr>
-<tr><td><b>inference_programme</b></td><td>What Nautilus achieves on mass_pix at f1e8 and f1e5</td><td>0</td><td>0</td><td>2026-09-04</td></tr>
 <tr><td><b>subhalo_validation</b></td><td>Does rectangular_adapt false-detect a subhalo on the clumpy source</td><td>0</td><td>5</td><td>2026-09-07</td></tr>
 </table>
 <h3>Nothing open</h3>
@@ -252,6 +251,9 @@ <h3>Nothing open</h3>
 <tr><td>slope_hierarchy</td><td>dormant</td><td>planned 2</td><td><button class="copy text" data-cmd="/cortex — retire the science project slope_hierarchy: confirm it is holding nothing live, then in PyAutoCortex run `python3 scripts/cortex.py retire slope_hierarchy --why &quot;&lt;one line on why&gt;&quot;`, `python3 scripts/cortex.py check`, `pyauto-brain cortex dashboard --apply`, and push the ledger. The row, its tasks and its rulings all stay — only the status and the note change." aria-label="Copy the retire prompt">📋 retire</button></td></tr>
 <tr><td>subhalo_simulations</td><td>dormant</td><td>none</td><td><button class="copy text" data-cmd="/cortex — retire the science project subhalo_simulations: confirm it is holding nothing live, then in PyAutoCortex run `python3 scripts/cortex.py retire subhalo_simulations --why &quot;&lt;one line on why&gt;&quot;`, `python3 scripts/cortex.py check`, `pyauto-brain cortex dashboard --apply`, and push the ledger. The row, its tasks and its rulings all stay — only the status and the note change." aria-label="Copy the retire prompt">📋 retire</button></td></tr>
 </table>
+<details><summary>1 retired</summary><ul>
+<li>inference_programme — retired 2026-09-07: restarting from scratch: the project setup was unsatisfactory and it informed how the Cortex manages projects; RAL outputs stashed at /mnt/ral/jnightin/inference_programme_retired_2026-09-07</li>
+</ul></details>
 <a id="projects"></a><h2>Projects <a class="mdsrc" href="https://github.com/PyAutoLabs/PyAutoCortex/blob/main/projects.yaml">markdown version</a></h2>
 <section class="project">
 <h3>euclid_dr1_prelim</h3>
@@ -262,20 +264,6 @@ <h3>euclid_dr1_prelim</h3>
 <p class="muted"><a href="https://github.com/PyAutoLabs/PyAutoCortex/tree/main/tasks/euclid_dr1_prelim/">tasks/euclid_dr1_prelim/</a></p>
 </section>
 <section class="project">
-<h3>inference_programme</h3>
-<p class="paths"><b>Local</b> <span class="pathchip">/home/jammy/Code/PyAutoLabs/autolens_profiling</span></p>
-<p class="paths"><b>Mirror</b> <span class="pathchip">/mnt/c/Users/Jammy/Science/inference_programme</span></p>
-<p class="paths"><b>RAL</b> <span class="pathchip">/mnt/ral/jnightin/autolens_profiling</span></p>
-<p class="muted">active · both partition · tasks: accepted 7 · dropped 11 · gated 1 · planned 4</p>
-<div class="task"><button class="copy" data-cmd="python3 scripts/cortex.py gates   # then, once they have closed: move tasks/inference_programme/nautilus_mass_pix_baseline.md ready
-# gates: autolens_profiling#218" aria-label="Copy the Claude command">📋</button><p><a href="https://github.com/PyAutoLabs/PyAutoCortex/blob/main/tasks/inference_programme/nautilus_mass_pix_baseline.md">What Nautilus achieves on mass_pix at f1e8 and f1e5</a><span class="tags"><span class="pill n">gated</span></span><span class="facets"> — autolens_profiling#218</span></p></div>
-<div class="task"><button class="copy" data-cmd="python3 scripts/cortex.py move tasks/inference_programme/cluster_extended_source_inference.md ready   # when the `Ready when:` clause in its `## Question` is met" aria-label="Copy the Claude command">📋</button><p><a href="https://github.com/PyAutoLabs/PyAutoCortex/blob/main/tasks/inference_programme/cluster_extended_source_inference.md">Cluster extended source inference testing gradient based fitting</a><span class="tags"><span class="pill g">planned</span></span></p></div>
-<div class="task"><button class="copy" data-cmd="python3 scripts/cortex.py move tasks/inference_programme/legacy_point_output_sweep.md ready   # when the `Ready when:` clause in its `## Question` is met" aria-label="Copy the Claude command">📋</button><p><a href="https://github.com/PyAutoLabs/PyAutoCortex/blob/main/tasks/inference_programme/legacy_point_output_sweep.md">Can spent point-source families be swept out of active output</a><span class="tags"><span class="pill g">planned</span></span></p></div>
-<div class="task"><button class="copy" data-cmd="python3 scripts/cortex.py move tasks/inference_programme/likelihood_term_levers.md ready   # when the `Ready when:` clause in its `## Question` is met" aria-label="Copy the Claude command">📋</button><p><a href="https://github.com/PyAutoLabs/PyAutoCortex/blob/main/tasks/inference_programme/likelihood_term_levers.md">Are likelihood terms behind gradient search failure on mass_pix</a><span class="tags"><span class="pill g">planned</span></span></p></div>
-<div class="task"><button class="copy" data-cmd="python3 scripts/cortex.py move tasks/inference_programme/prodigy_mass_pix.md ready   # when the `Ready when:` clause in its `## Question` is met" aria-label="Copy the Claude command">📋</button><p><a href="https://github.com/PyAutoLabs/PyAutoCortex/blob/main/tasks/inference_programme/prodigy_mass_pix.md">Does a gradient search beat Nautilus on mass_pix</a><span class="tags"><span class="pill g">planned</span></span></p></div>
-<p class="muted"><a href="https://github.com/PyAutoLabs/PyAutoCortex/tree/main/tasks/inference_programme/">tasks/inference_programme/</a> · <a href="https://github.com/PyAutoLabs/autolens_profiling/issues">issues</a></p>
-</section>
-<section class="project">
 <h3>subhalo_validation</h3>
 <p class="paths"><b>Local</b> <span class="pathchip">/mnt/c/Users/Jammy/Science/subhalo_validation</span></p>
 <p class="paths"><b>RAL</b> <span class="pathchip">/mnt/ral/jnightin/subhalo_validation</span></p>
@@ -292,6 +280,11 @@ <h3>subhalo_validation</h3>
 <a id="recent-rulings"></a><h2>Recent rulings <a class="mdsrc" href="https://github.com/PyAutoLabs/PyAutoCortex/tree/main/rulings/">markdown version</a></h2><p class="muted">The ledger of record, newest first. A verdict recorded only outside the Cortex does not exist.</p>
 <table class="map">
 <tr><th>Ruling</th><th>Verb</th><th>Task</th><th>Batch</th></tr>
+<tr><td><a href="https://github.com/PyAutoLabs/PyAutoCortex/blob/main/rulings/2026/09/R-20260907-06.md">R-20260907-06</a></td><td>drop</td><td>tasks/inference_programme/legacy_point_output_sweep.md</td><td>-</td></tr>
+<tr><td><a href="https://github.com/PyAutoLabs/PyAutoCortex/blob/main/rulings/2026/09/R-20260907-05.md">R-20260907-05</a></td><td>drop</td><td>tasks/inference_programme/cluster_extended_source_inference.md</td><td>-</td></tr>
+<tr><td><a href="https://github.com/PyAutoLabs/PyAutoCortex/blob/main/rulings/2026/09/R-20260907-04.md">R-20260907-04</a></td><td>drop</td><td>tasks/inference_programme/likelihood_term_levers.md</td><td>-</td></tr>
+<tr><td><a href="https://github.com/PyAutoLabs/PyAutoCortex/blob/main/rulings/2026/09/R-20260907-03.md">R-20260907-03</a></td><td>drop</td><td>tasks/inference_programme/prodigy_mass_pix.md</td><td>-</td></tr>
+<tr><td><a href="https://github.com/PyAutoLabs/PyAutoCortex/blob/main/rulings/2026/09/R-20260907-02.md">R-20260907-02</a></td><td>drop</td><td>tasks/inference_programme/nautilus_mass_pix_baseline.md</td><td>-</td></tr>
 <tr><td><a href="https://github.com/PyAutoLabs/PyAutoCortex/blob/main/rulings/2026/09/R-20260907-01.md">R-20260907-01</a></td><td>accept</td><td>tasks/subhalo_validation/delaunay_adapt_split_pl_sersic_0.md</td><td>-</td></tr>
 <tr><td><a href="https://github.com/PyAutoLabs/PyAutoCortex/blob/main/rulings/2026/09/R-20260904-05.md">R-20260904-05</a></td><td>drop</td><td>tasks/inference_programme/multiband_compile_census_completion.md</td><td>-</td></tr>
 <tr><td><a href="https://github.com/PyAutoLabs/PyAutoCortex/blob/main/rulings/2026/09/R-20260904-04.md">R-20260904-04</a></td><td>drop</td><td>tasks/inference_programme/mass_pix_gradient_cost_probe.md</td><td>-</td></tr>
@@ -299,11 +292,6 @@ <h3>subhalo_validation</h3>
 <tr><td><a href="https://github.com/PyAutoLabs/PyAutoCortex/blob/main/rulings/2026/09/R-20260904-02.md">R-20260904-02</a></td><td>drop</td><td>tasks/inference_programme/phase5_mesh_gradient_positions_on.md</td><td>-</td></tr>
 <tr><td><a href="https://github.com/PyAutoLabs/PyAutoCortex/blob/main/rulings/2026/09/R-20260904-01.md">R-20260904-01</a></td><td>accept</td><td>tasks/inference_programme/refs_v1_positions_on_completion.md</td><td>-</td></tr>
 <tr><td><a href="https://github.com/PyAutoLabs/PyAutoCortex/blob/main/rulings/2026/09/R-20260902-10.md">R-20260902-10</a></td><td>accept</td><td>tasks/inference_programme/phase4_positions_gate_b2_reuse.md</td><td>2026-09-02-pm</td></tr>
-<tr><td><a href="https://github.com/PyAutoLabs/PyAutoCortex/blob/main/rulings/2026/09/R-20260902-09.md">R-20260902-09</a></td><td>accept</td><td>tasks/inference_programme/phase3_prodigy_reliability_gate_b1_reuse.md</td><td>2026-09-02-pm</td></tr>
-<tr><td><a href="https://github.com/PyAutoLabs/PyAutoCortex/blob/main/rulings/2026/09/R-20260902-08.md">R-20260902-08</a></td><td>accept</td><td>tasks/inference_programme/phase2_nss_mainline_gate_a_reuse.md</td><td>2026-09-02-pm</td></tr>
-<tr><td><a href="https://github.com/PyAutoLabs/PyAutoCortex/blob/main/rulings/2026/09/R-20260902-07.md">R-20260902-07</a></td><td>rerun</td><td>tasks/subhalo_validation/delaunay_adapt_split_pl_sersic_0.md</td><td>-</td></tr>
-<tr><td><a href="https://github.com/PyAutoLabs/PyAutoCortex/blob/main/rulings/2026/09/R-20260902-06.md">R-20260902-06</a></td><td>rerun</td><td>tasks/subhalo_validation/rectangular_adapt_pl_sersic_0.md</td><td>-</td></tr>
-<tr><td><a href="https://github.com/PyAutoLabs/PyAutoCortex/blob/main/rulings/2026/09/R-20260902-05.md">R-20260902-05</a></td><td>rerun</td><td>tasks/subhalo_validation/delaunay_adapt_split_pl_eff_0.md</td><td>-</td></tr>
 </table>
 <ul class="boards"><li class="muted" style="padding:.3rem 0">Boards:</li><li><a data-organ="brain" href="https://pyautolabs.github.io/PyAutoBrain/">Brain</a></li><li><a data-organ="mind" href="https://pyautolabs.github.io/PyAutoMind/">Mind</a></li><li><a data-organ="memory" href="https://pyautolabs.github.io/PyAutoMemory/">Memory</a></li><li><a data-organ="heart" href="https://pyautolabs.github.io/PyAutoHeart/">Heart</a></li><li><a data-organ="hands" href="https://pyautolabs.github.io/PyAutoHands/">Hands</a></li><li><a data-organ="organism" href="https://pyautolabs.github.io/PyAutoScientist/">Scientist</a></li></ul>
 <script>async function copyCmd(b){
diff --git a/dashboard.md b/dashboard.md
index 226719d..854f66c 100644
--- a/dashboard.md
+++ b/dashboard.md
@@ -11,7 +11,7 @@ Every science task the Cortex is holding, on one page: one line each under the p
 | Where | Count |
 |-------|------:|
 | [Awaiting ruling](#awaiting-ruling) | 0 |
-| [Recent rulings](#recent-rulings) | 28 |
+| [Recent rulings](#recent-rulings) | 33 |
 
 <details><summary>📋 check in since last time</summary>
 
@@ -30,7 +30,6 @@ One row per active project — the next thing, what it is holding, and when it l
 | Project | Next task | Awaiting | Running | Last ruling |
 |---|---|---:|---:|---|
 | euclid_dr1_prelim | Ten real DR1 lenses fitted from the pipeline repo alone | 0 | 1 | - |
-| inference_programme | What Nautilus achieves on mass_pix at f1e8 and f1e5 | 0 | 0 | 2026-09-04 |
 | subhalo_validation | Does rectangular_adapt false-detect a subhalo on the clumpy source | 0 | 5 | 2026-09-07 |
 
 #### Nothing open
@@ -110,6 +109,12 @@ One row per active project — the next thing, what it is holding, and when it l
 
 </details>
 
+<details><summary>1 retired</summary>
+
+- inference_programme — retired 2026-09-07: restarting from scratch: the project setup was unsatisfactory and it informed how the Cortex manages projects; RAL outputs stashed at /mnt/ral/jnightin/inference_programme_retired_2026-09-07
+
+</details>
+
 ## Projects
 
 [markdown version](https://github.com/PyAutoLabs/PyAutoCortex/blob/main/projects.yaml)
@@ -132,59 +137,6 @@ Report where the runs of the PyAutoCortex task tasks/euclid_dr1_prelim/dr1_preli
 
 [tasks/euclid_dr1_prelim/](https://github.com/PyAutoLabs/PyAutoCortex/tree/main/tasks/euclid_dr1_prelim/)
 
-### inference_programme
-
-**Local** `/home/jammy/Code/PyAutoLabs/autolens_profiling`
-
-**Mirror** `/mnt/c/Users/Jammy/Science/inference_programme`
-
-**RAL** `/mnt/ral/jnightin/autolens_profiling`
-
-active · both partition · tasks: accepted 7 · dropped 11 · gated 1 · planned 4
-
-<details><summary>📋 <a href="tasks/inference_programme/nautilus_mass_pix_baseline.md">What Nautilus achieves on mass_pix at f1e8 and f1e5</a> — <b>gated</b> — autolens_profiling#218</summary>
-
-```
-python3 scripts/cortex.py gates   # then, once they have closed: move tasks/inference_programme/nautilus_mass_pix_baseline.md ready
-# gates: autolens_profiling#218
-```
-
-</details>
-
-<details><summary>📋 <a href="tasks/inference_programme/cluster_extended_source_inference.md">Cluster extended source inference testing gradient based fitting</a> — <b>planned</b></summary>
-
-```
-python3 scripts/cortex.py move tasks/inference_programme/cluster_extended_source_inference.md ready   # when the `Ready when:` clause in its `## Question` is met
-```
-
-</details>
-
-<details><summary>📋 <a href="tasks/inference_programme/legacy_point_output_sweep.md">Can spent point-source families be swept out of active output</a> — <b>planned</b></summary>
-
-```
-python3 scripts/cortex.py move tasks/inference_programme/legacy_point_output_sweep.md ready   # when the `Ready when:` clause in its `## Question` is met
-```
-
-</details>
-
-<details><summary>📋 <a href="tasks/inference_programme/likelihood_term_levers.md">Are likelihood terms behind gradient search failure on mass_pix</a> — <b>planned</b></summary>
-
-```
-python3 scripts/cortex.py move tasks/inference_programme/likelihood_term_levers.md ready   # when the `Ready when:` clause in its `## Question` is met
-```
-
-</details>
-
-<details><summary>📋 <a href="tasks/inference_programme/prodigy_mass_pix.md">Does a gradient search beat Nautilus on mass_pix</a> — <b>planned</b></summary>
-
-```
-python3 scripts/cortex.py move tasks/inference_programme/prodigy_mass_pix.md ready   # when the `Ready when:` clause in its `## Question` is met
-```
-
-</details>
-
-[tasks/inference_programme/](https://github.com/PyAutoLabs/PyAutoCortex/tree/main/tasks/inference_programme/) · [issues](https://github.com/PyAutoLabs/autolens_profiling/issues)
-
 ### subhalo_validation
 
 **Local** `/mnt/c/Users/Jammy/Science/subhalo_validation`
@@ -247,6 +199,11 @@ cd /mnt/c/Users/Jammy/Science/subhalo_validation && hpc/sync jobs
 
 | Ruling | Verb | Task | Batch |
 |---|---|---|---|
+| [R-20260907-06](rulings/2026/09/R-20260907-06.md) | drop | tasks/inference_programme/legacy_point_output_sweep.md | - |
+| [R-20260907-05](rulings/2026/09/R-20260907-05.md) | drop | tasks/inference_programme/cluster_extended_source_inference.md | - |
+| [R-20260907-04](rulings/2026/09/R-20260907-04.md) | drop | tasks/inference_programme/likelihood_term_levers.md | - |
+| [R-20260907-03](rulings/2026/09/R-20260907-03.md) | drop | tasks/inference_programme/prodigy_mass_pix.md | - |
+| [R-20260907-02](rulings/2026/09/R-20260907-02.md) | drop | tasks/inference_programme/nautilus_mass_pix_baseline.md | - |
 | [R-20260907-01](rulings/2026/09/R-20260907-01.md) | accept | tasks/subhalo_validation/delaunay_adapt_split_pl_sersic_0.md | - |
 | [R-20260904-05](rulings/2026/09/R-20260904-05.md) | drop | tasks/inference_programme/multiband_compile_census_completion.md | - |
 | [R-20260904-04](rulings/2026/09/R-20260904-04.md) | drop | tasks/inference_programme/mass_pix_gradient_cost_probe.md | - |
@@ -254,11 +211,6 @@ cd /mnt/c/Users/Jammy/Science/subhalo_validation && hpc/sync jobs
 | [R-20260904-02](rulings/2026/09/R-20260904-02.md) | drop | tasks/inference_programme/phase5_mesh_gradient_positions_on.md | - |
 | [R-20260904-01](rulings/2026/09/R-20260904-01.md) | accept | tasks/inference_programme/refs_v1_positions_on_completion.md | - |
 | [R-20260902-10](rulings/2026/09/R-20260902-10.md) | accept | tasks/inference_programme/phase4_positions_gate_b2_reuse.md | 2026-09-02-pm |
-| [R-20260902-09](rulings/2026/09/R-20260902-09.md) | accept | tasks/inference_programme/phase3_prodigy_reliability_gate_b1_reuse.md | 2026-09-02-pm |
-| [R-20260902-08](rulings/2026/09/R-20260902-08.md) | accept | tasks/inference_programme/phase2_nss_mainline_gate_a_reuse.md | 2026-09-02-pm |
-| [R-20260902-07](rulings/2026/09/R-20260902-07.md) | rerun | tasks/subhalo_validation/delaunay_adapt_split_pl_sersic_0.md | - |
-| [R-20260902-06](rulings/2026/09/R-20260902-06.md) | rerun | tasks/subhalo_validation/rectangular_adapt_pl_sersic_0.md | - |
-| [R-20260902-05](rulings/2026/09/R-20260902-05.md) | rerun | tasks/subhalo_validation/delaunay_adapt_split_pl_eff_0.md | - |
 
 ---
 
diff --git a/projects.yaml b/projects.yaml
index b611d38..0791ccc 100644
--- a/projects.yaml
+++ b/projects.yaml
@@ -52,8 +52,8 @@ inference_programme:
   ledger: results/notes/inference/PROGRAMME.md
   witness_file: results/searches/**/*.json
   partition: both
-  status: active
-  note: "DECISIONS.md is commentary; pull-only CLI — RAL is a git checkout; gpu-1 excluded"
+  status: retired
+  note: "retired 2026-09-07: restarting from scratch: the project setup was unsatisfactory and it informed how the Cortex manages projects; RAL outputs stashed at /mnt/ral/jnightin/inference_programme_retired_2026-09-07"
 
 subhalo_validation:
   remote: PyAutoLabs/subhalo_validation
diff --git a/rulings/2026/09/R-20260907-02.md b/rulings/2026/09/R-20260907-02.md
new file mode 100644
index 0000000..14eeea2
--- /dev/null
+++ b/rulings/2026/09/R-20260907-02.md
@@ -0,0 +1,21 @@
+# R-20260907-02 — drop inference_programme nautilus_mass_pix_baseline
+
+Project: inference_programme
+Task: tasks/inference_programme/nautilus_mass_pix_baseline.md
+Runs:
+Ruling: drop
+Reviewed-at: 2026-09-07T21:03Z
+
+## Ruling
+
+I want to retire inference_programme fully, I will start it again from scratch, I am not happy with how i set up that science project and it has done lots to inform how Cortex manages things.
+
+## Evidence
+
+- `autolens_profiling/results/notes/gradient_slam/LEDGER.md` — the `mass_pix` target definition and the provenance of every fixed value
+- `autolens_profiling/scripts/imaging/searches/nautilus/mass_pix.py` — the driver (written by the dev leg)
+- `autolens_profiling/hpc/batch_gpu/` — the two-arm submit script and its `WALL-BASIS` block
+- `inference_programme` (project row): `output/searches/nautilus/imaging/mass_pix/hst/pos_tauto0.2_f1e8/` and `.../pos_tauto0.2_f1e5/` — the two run trees, each with `positions.info`
+- `autolens_profiling/results/searches/nautilus/imaging/mass_pix/hst/` — the result rows
+- `logs/output/` and `logs/error/` on the mirror
+- `autolens_profiling/results/baselines/InferenceRefs_v1/pixelization_pos_fp64/` — the nearest certified neighbour (12 free parameters vs this cell's 7)
diff --git a/rulings/2026/09/R-20260907-03.md b/rulings/2026/09/R-20260907-03.md
new file mode 100644
index 0000000..d31e366
--- /dev/null
+++ b/rulings/2026/09/R-20260907-03.md
@@ -0,0 +1,22 @@
+# R-20260907-03 — drop inference_programme prodigy_mass_pix
+
+Project: inference_programme
+Task: tasks/inference_programme/prodigy_mass_pix.md
+Runs:
+Ruling: drop
+Reviewed-at: 2026-09-07T21:03Z
+
+## Ruling
+
+I want to retire inference_programme fully, I will start it again from scratch, I am not happy with how i set up that science project and it has done lots to inform how Cortex manages things.
+
+## Evidence
+
+- `autolens_profiling/results/notes/gradient_slam/LEDGER.md` — the question, the inherited MGE Prodigy evidence and the engine-split positions rule
+- `autolens_profiling/scripts/imaging/searches/multi_start_prodigy_autoconv/mass_pix.py` — the driver (written by the dev leg)
+- `autolens_profiling/scripts/misc/searches/_samplers.py` — the per-cell `imaging:mass_pix` rows (n_starts 16, batch 4, n_steps 3000)
+- `autolens_profiling/hpc/batch_gpu/` — the seeds 0–4 array submit and its `WALL-BASIS` block
+- `inference_programme` (project row): `output/searches/multi_start_prodigy_autoconv/imaging/mass_pix/hst/pos_tauto0.2_f1e5/` — the five run trees, each with `positions.info`
+- `autolens_profiling/results/searches/multi_start_prodigy_autoconv/imaging/mass_pix/hst/` — the result rows
+- `logs/output/` and `logs/error/` on the mirror
+- `nautilus_mass_pix_baseline` — the bar this is scored against
diff --git a/rulings/2026/09/R-20260907-04.md b/rulings/2026/09/R-20260907-04.md
new file mode 100644
index 0000000..0194d96
--- /dev/null
+++ b/rulings/2026/09/R-20260907-04.md
@@ -0,0 +1,21 @@
+# R-20260907-04 — drop inference_programme likelihood_term_levers
+
+Project: inference_programme
+Task: tasks/inference_programme/likelihood_term_levers.md
+Runs:
+Ruling: drop
+Reviewed-at: 2026-09-07T21:03Z
+
+## Ruling
+
+I want to retire inference_programme fully, I will start it again from scratch, I am not happy with how i set up that science project and it has done lots to inform how Cortex manages things.
+
+## Evidence
+
+- `autolens_profiling/results/notes/gradient_slam/LEDGER.md` — the phase table and the inherited rules
+- `prodigy_mass_pix` — the control this task is paired against; its ruling decides whether this arms at all
+- `autolens_profiling#166` — the slogdet-as-PyAutoArray-default reminder this lever folds
+- `autolens_profiling/results/notes/inference/phase_08_regularization/RESULTS.md` — the retired programme's slogdet A/B, for what was already learned
+- `PyAutoArray` `constant.py:53` and the `adapt.py` regularization schemes — the absolute `1e-8` lift vs `jitter_relative`
+- `inference_programme` (project row): `output/searches/multi_start_prodigy_autoconv/imaging/mass_pix/hst/` — the lever arms land beside the phase 22 control
+- `logs/output/` and `logs/error/` on the mirror
diff --git a/rulings/2026/09/R-20260907-05.md b/rulings/2026/09/R-20260907-05.md
new file mode 100644
index 0000000..a1f7a73
--- /dev/null
+++ b/rulings/2026/09/R-20260907-05.md
@@ -0,0 +1,17 @@
+# R-20260907-05 — drop inference_programme cluster_extended_source_inference
+
+Project: inference_programme
+Task: tasks/inference_programme/cluster_extended_source_inference.md
+Runs:
+Ruling: drop
+Reviewed-at: 2026-09-07T21:03Z
+
+## Ruling
+
+I want to retire inference_programme fully, I will start it again from scratch, I am not happy with how i set up that science project and it has done lots to inform how Cortex manages things.
+
+## Evidence
+
+- Mind `draft/feature/autolens/source_cluster_arc.md` — the arc ledger (this is its line 11)
+- Mind `draft/feature/workspaces/cluster_pixelized_analysisfactor.md` — cluster phase 10, the gate
+- `inference_programme` (project row): `output/searches/*/cluster/` — the cluster trees the
diff --git a/rulings/2026/09/R-20260907-06.md b/rulings/2026/09/R-20260907-06.md
new file mode 100644
index 0000000..ed169ef
--- /dev/null
+++ b/rulings/2026/09/R-20260907-06.md
@@ -0,0 +1,19 @@
+# R-20260907-06 — drop inference_programme legacy_point_output_sweep
+
+Project: inference_programme
+Task: tasks/inference_programme/legacy_point_output_sweep.md
+Runs:
+Ruling: drop
+Reviewed-at: 2026-09-07T21:03Z
+
+## Ruling
+
+I want to retire inference_programme fully, I will start it again from scratch, I am not happy with how i set up that science project and it has done lots to inform how Cortex manages things.
+
+## Evidence
+
+- `/mnt/ral/jnightin/autolens_profiling/output/` — the three point-source
+- `/mnt/c/Users/Jammy/Science/inference_programme/`
+- `autolens_profiling/results/notes/inference/DECISIONS.md` (2026-09-01 entry —
+- `autolens_profiling/results/baselines/InferenceRefs_v1/SUBMIT_LIST.md` — the
+- `rulings/2026/09/R-20260902-01.md`, `rulings/2026/09/R-20260904-01.md`
diff --git a/tasks/inference_programme/cluster_extended_source_inference.md b/tasks/inference_programme/cluster_extended_source_inference.md
index b94b625..ecf480a 100644
--- a/tasks/inference_programme/cluster_extended_source_inference.md
+++ b/tasks/inference_programme/cluster_extended_source_inference.md
@@ -2,12 +2,12 @@
 
 Project: inference_programme
 Summary: Cluster extended source inference testing gradient based fitting
-State: planned
+State: dropped
 Gates:
 Witness: a written feasibility verdict with profiling numbers and a go/no-go for a follow-up implementation prompt — not shipped inference machinery
 Budget: 24:00
 Runs:
-Ruling:
+Ruling: R-20260907-05
 Review-minutes: 25
 Epic: cluster-strong-lensing
 Filed: 2026-08-19
@@ -49,4 +49,4 @@ implementation prompt — not shipped inference machinery.
 
 ## Ruling
 
-(none)
+R-20260907-05 — drop
diff --git a/tasks/inference_programme/legacy_point_output_sweep.md b/tasks/inference_programme/legacy_point_output_sweep.md
index bb4d57a..7caf04e 100644
--- a/tasks/inference_programme/legacy_point_output_sweep.md
+++ b/tasks/inference_programme/legacy_point_output_sweep.md
@@ -2,12 +2,12 @@
 
 Project: inference_programme
 Summary: Can spent point-source families be swept out of active output
-State: planned
+State: dropped
 Gates: autolens_profiling#205
 Witness: `output/legacy_point/` exists on RAL and on the laptop mirror holding all three point-source families (`image_plane`/`source_plane` fits, `Cluster`, `point_source`) with their directory structure preserved; the active `output/` tree no longer contains them; the certified `InferenceRefs_v1` mesh rows (R-20260902-01, R-20260904-01) are STILL PRESENT in the active tree and were not moved; a spot check finds any moved run under its new path; nothing deleted; a dated one-line note appended beside the ledger
 Budget: 2:00
 Runs:
-Ruling:
+Ruling: R-20260907-06
 Review-minutes: 5
 Filed: 2026-09-01
 Migrated-from: PyAutoMind/draft/maintenance/autolens_profiling/legacy_point_output_sweep.md
@@ -113,3 +113,5 @@ actually cite.
 ## Runs
 
 ## Ruling
+
+R-20260907-06 — drop
diff --git a/tasks/inference_programme/likelihood_term_levers.md b/tasks/inference_programme/likelihood_term_levers.md
index b54b0a0..d1e72f6 100644
--- a/tasks/inference_programme/likelihood_term_levers.md
+++ b/tasks/inference_programme/likelihood_term_levers.md
@@ -2,12 +2,12 @@
 
 Project: inference_programme
 Summary: Are likelihood terms behind gradient search failure on mass_pix
-State: planned
+State: dropped
 Gates:
 Witness: each lever arm runs the same `mass_pix` gradient configuration as phase 22 and delivers on its own artefacts — `.completed`, `positions.info`, `.err` free of Tracebacks, `.out` ending "Finished.", a result row whose `target_id` recomputes from `_targets.py`; and each arm reports, against the phase 22 control it is paired with, the change in NaN-lane count, the change in p_hit against phase 21's `f1e5` max logL, and the change in wall — so that a lever is judged on whether it removes the failure phase 22 measured, not on whether it changes a number
 Budget: 4:00
 Runs:
-Ruling:
+Ruling: R-20260907-04
 Review-minutes: 20
 Epic: gradient-slam-baseline
 Filed: 2026-09-04
@@ -56,4 +56,4 @@ each lever arm runs the same `mass_pix` gradient configuration as phase 22 and d
 
 ## Ruling
 
-(none)
+R-20260907-04 — drop
diff --git a/tasks/inference_programme/nautilus_mass_pix_baseline.md b/tasks/inference_programme/nautilus_mass_pix_baseline.md
index 68d2aa5..21f3c66 100644
--- a/tasks/inference_programme/nautilus_mass_pix_baseline.md
+++ b/tasks/inference_programme/nautilus_mass_pix_baseline.md
@@ -2,12 +2,12 @@
 
 Project: inference_programme
 Summary: What Nautilus achieves on mass_pix at f1e8 and f1e5
-State: gated
+State: dropped
 Gates: autolens_profiling#218
 Witness: both Nautilus arms deliver on their own artefacts — a `.completed` marker and a `positions.info` file in each run dir; `.err` free of Tracebacks; `.out` ends "Finished." with zero "Fit Already Completed"; a result row per arm carrying a `target_id` that recomputes from `_targets.py`; no overflow signature (no finite `log_l` above the Fitness ceiling); each arm physical — `einstein_radius` recovered at 1.60 ± 0.02 with `|ell_comps| < 1`; and max logL, log evidence, likelihood evals and sampler wall recorded for both arms
 Budget: 4:00
 Runs:
-Ruling:
+Ruling: R-20260907-02
 Review-minutes: 15
 Epic: gradient-slam-baseline
 Filed: 2026-09-04
@@ -77,4 +77,4 @@ both Nautilus arms deliver on their own artefacts — a `.completed` marker and
 
 ## Ruling
 
-(none)
+R-20260907-02 — drop
diff --git a/tasks/inference_programme/prodigy_mass_pix.md b/tasks/inference_programme/prodigy_mass_pix.md
index 0f68768..86838fc 100644
--- a/tasks/inference_programme/prodigy_mass_pix.md
+++ b/tasks/inference_programme/prodigy_mass_pix.md
@@ -2,12 +2,12 @@
 
 Project: inference_programme
 Summary: Does a gradient search beat Nautilus on mass_pix
-State: planned
+State: dropped
 Gates:
 Witness: five `MultiStartProdigy` autoconv seeds (0–4) on the `mass_pix` cell at `pos_tauto0.2_f1e5` each deliver on their own artefacts — `.completed`, `positions.info`, `.err` free of Tracebacks, `.out` ending "Finished.", a result row whose `target_id` recomputes from `_targets.py`, no overflow signature; a seed HITS when its best log-posterior lands within 2 nats of phase 21's `pos_tauto0.2_f1e5` max logL AND its best point is physical (`|ell_comps| < 1`, `einstein_radius` 1.60 ± 0.02); p_hit over the five seeds is reported with its Wilson-95 lower bound, and the per-seed wall is reported against phase 21's sampler wall on the same arm; a lane that ends NaN or pinned is counted and reported, never silently dropped
 Budget: 4:00
 Runs:
-Ruling:
+Ruling: R-20260907-03
 Review-minutes: 25
 Epic: gradient-slam-baseline
 Filed: 2026-09-04
@@ -62,4 +62,4 @@ five `MultiStartProdigy` autoconv seeds (0–4) on the `mass_pix` cell at `pos_t
 
 ## Ruling
 
-(none)
+R-20260907-03 — drop