diff --git a/complete/2026/08/jax-traceback-filtering-release-harness.md b/complete/2026/08/jax-traceback-filtering-release-harness.md
new file mode 100644
index 00000000..f76d5899
--- /dev/null
+++ b/complete/2026/08/jax-traceback-filtering-release-harness.md
@@ -0,0 +1,178 @@
+# Release/smoke script harness stopped filtering JAX tracebacks
+
+Issue: PyAutoLabs/PyAutoHeart#186 (CLOSED 2026-08-27)
+
+## What was wrong
+
+JAX removes its own frames from a traceback by default and, in practice, takes
+the exception that names the failure with them. A workspace script failing under
+JAX inside Heart's script harness reported only:
+
+```
+scripts/interferometer/start_here.py ... FAIL (19.5s)
+For simplicity, JAX has removed its internal frames from the traceback of the
+following exception. Set JAX_TRACEBACK_FILTERING=off to include these.
+```
+
+Which script failed, and nothing about why.
+
+The cost was measured, not hypothetical. Heart run 30516167217 (2026-07-30)
+failed **both** interferometer shards (`autogalaxy` and `autolens`) with exactly
+that placeholder, so that night read as an unidentified failure. The real cause —
+`jax.errors.JaxRuntimeError: RESOURCE_EXHAUSTED: Out of memory allocating
+85898814480 bytes` — became legible only in run 30607596240 the following night,
+where the message happened to survive. One night was spent identifying a failure
+the harness already had the information to name.
+
+## What shipped
+
+| Repo | PR | Merged | Change |
+|---|---|---|---|
+| PyAutoHeart | [#187](https://github.com/PyAutoLabs/PyAutoHeart/pull/187) | 2026-08-27 (`c8b093e1`) | `JAX_TRACEBACK_FILTERING: "off"` — workflow-level `env:` in `workspace-validation.yml`, and the workspace-runner step's `env:` in `smoke-tests.yml`. 2 files, +18/−0 |
+
+Workflow-level `env:` reaches every job and step of `workspace-validation.yml`,
+so both `mode=smoke` and `mode=release`, and every project in the matrix, inherit
+it. `run_python.py`'s per-script env profiles inherit it too — a profile only
+*sets* the keys it is given and never clears inherited vars, which is the
+behaviour `profile_release.yaml`'s own header already relies on.
+`workspace-smoke.yml` needed no edit: it is a thin caller of
+`workspace-validation.yml`.
+
+CI before merge: Heart Tests run 33093776599, `pytest (3.12)` + `pytest (3.13)`
+both success, `mergeable_state: clean`. Locally `python3 -m pytest -q -n auto` →
+641 passed.
+
+## The exit criterion is deferred, not met
+
+The issue asked that a real JAX exception print its own type and message in a job
+log, "verified against a real (or deliberately induced) JAX failure in a CI run,
+not only by reading the diff". Nothing has failed under JAX since the merge, so
+that has not been observed. The next JAX failure in `run_scripts` or the smoke
+runner is the proof. This is recorded as shipped-but-unproven on purpose: the
+whole defect is that the diff looks obviously right and the log is what has to
+change.
+
+## Traps worth keeping
+
+- **`off` must be quoted in YAML.** Bare `off` is a YAML 1.1 boolean, so
+ `JAX_TRACEBACK_FILTERING: off` hands GitHub Actions `false`, not the string JAX
+ expects. Checked with `yaml.safe_load` rather than assumed.
+- **A stale prompt can outlive its own fix by a month.** This task exists only
+ because `/start_dev` on
+ `draft/bug/autolens/interferometer_release_leg_oom.md` found the prompt's
+ headline OOM already shipped —
+ `complete/2026/08/interferometer-start-here-integrate-oom.md`, 2026-07-31,
+ `MultiStartProdigy` running 48 starts as one unbatched `vmap`. That prompt sat
+ in `draft/` as pickable backlog for 27 days, and its triage notes pointed at
+ the wrong subsystem (NUFFT/transformer). It was retired to
+ `complete/archive/shelved/` in the same run. **Check `complete/` for the slug
+ before planning a bug prompt**, not after.
+- **A prompt's own amendment can be the only live part of it.** The retired
+ prompt's 2026-08-04 amendment carried two consequences. One ("it is not
+ autolens-specific") was already closed — both workspaces carry `batch_size=4`.
+ The other was this task. Reading only the headline would have lost it.
+
+## Original prompt
+
+`active/jax_traceback_filtering_release_harness.md`, filed and issued
+2026-08-27 by the same `/start_dev` run that retired its parent.
+
+## Not done
+
+12 workspace scripts still call `MultiStartProdigy` with no `batch_size`
+(autolens: `group/`, `point_source/`, `multi_dataset/`, `cluster/`, `weak/`, …;
+autogalaxy: `multi_galaxy/modeling.py`, `multi_dataset/start_here.py`,
+`cluster/modeling.py`). Cheaper datasets than a 108k-visibility interferometer
+fit, so likely fine — surfaced during this run, deliberately not filed.
+
+## Original prompt
+
+# Release/smoke script harness filters JAX tracebacks away, hiding the real exception
+
+Type: bug
+Target: ci
+Repos:
+- @PyAutoHeart
+Difficulty: small
+Autonomy: safe
+Priority: medium
+Status: formalised
+Filed: 2026-08-27
+Issued: 2026-08-27
+
+## Symptom
+
+When a workspace script fails under JAX inside PyAutoHeart's script harness
+(`.github/workflows/workspace-validation.yml`, job `run_scripts`, both the
+`smoke` and `release` legs), JAX's default traceback filter removes its own
+internal frames *and*, in practice, the exception that matters:
+
+```
+scripts/interferometer/start_here.py ... FAIL (19.5s)
+For simplicity, JAX has removed its internal frames from the traceback of the
+following exception. Set JAX_TRACEBACK_FILTERING=off to include these.
+```
+
+That is the whole failure record. The reader is told which script failed and
+nothing about why.
+
+## Why it matters — the cost is measured, not hypothetical
+
+PyAutoHeart run 30516167217 (2026-07-30) failed both interferometer shards
+(`autogalaxy` and `autolens`) with exactly the message above. Because the real
+exception was filtered away, that night read as an unidentified failure. The
+same defect surfaced its true cause only the following night, run
+30607596240 (2026-07-31), where the message survived intact:
+
+```
+jax.errors.JaxRuntimeError: RESOURCE_EXHAUSTED: Out of memory allocating 85898814480 bytes.
+```
+
+One night was spent identifying a failure the harness already had the
+information to name. The underlying OOM was root-caused and fixed the same day
+(`autolens_workspace#450`; record:
+`complete/2026/08/interferometer-start-here-integrate-oom.md`) — this task is
+about the *harness*, which still filters, and will do the same to the next
+JAX failure.
+
+A gate whose error messages are filtered away is the same category of problem
+as the exit-code contract fixed in PyAutoBrain#196: the signal exists, the
+harness discards it before a human sees it.
+
+## Current state (verified 2026-08-27)
+
+`JAX_TRACEBACK_FILTERING` appears nowhere in PyAutoHeart:
+
+```
+$ grep -rn "JAX_TRACEBACK_FILTERING" . # PyAutoHeart @ d576003 — no matches
+```
+
+Nor in the workspaces' `config/build/profile_release.yaml` /
+`profile_smoke.yaml` defaults, both of which already pin every JAX-adjacent var
+they care about (`JAX_ENABLE_X64`, `PYAUTO_DISABLE_JAX`, …) explicitly.
+
+## Proposed fix
+
+Set `JAX_TRACEBACK_FILTERING: "off"` once, in the workflow-level `env:` block of
+`PyAutoHeart/.github/workflows/workspace-validation.yml` — the block that
+already carries `PYAUTO_SKIP_WORKSPACE_VERSION_CHECK`. One place covers both
+the `smoke` and `release` legs and every workspace in the matrix, and
+`run_python.py`'s per-script env profiles inherit it (a profile only *sets* the
+keys it is given; it never clears inherited vars).
+
+Consider whether the same belongs in the other script-running workflows
+(`smoke-tests.yml`, `workspace-smoke.yml`) for the same reason.
+
+## Exit criteria
+
+- A JAX exception raised inside a workspace script under `run_scripts` reports
+ its own exception type and message in the job log, not the
+ "JAX has removed its internal frames" placeholder.
+- Verified against a real (or deliberately induced) JAX failure in a CI run,
+ not only by reading the diff.
+
+## Origin
+
+Split out of `draft/bug/autolens/interferometer_release_leg_oom.md` (its
+2026-08-04 amendment, consequence 2) when that prompt was retired as superseded
+— the OOM it described had already shipped, this harness gap had not.
diff --git a/draft/bug/autolens/interferometer_release_leg_oom.md b/complete/archive/shelved/interferometer_release_leg_oom.md
similarity index 71%
rename from draft/bug/autolens/interferometer_release_leg_oom.md
rename to complete/archive/shelved/interferometer_release_leg_oom.md
index 17e5d642..da630395 100644
--- a/draft/bug/autolens/interferometer_release_leg_oom.md
+++ b/complete/archive/shelved/interferometer_release_leg_oom.md
@@ -1,3 +1,28 @@
+> **RETIRED 2026-08-27 — superseded, not actioned as written.**
+>
+> The OOM this prompt describes was root-caused and fixed before the prompt's own
+> amendment was written. See `complete/2026/08/interferometer-start-here-integrate-oom.md`:
+> `MultiStartProdigy` ran 48 starts as a single unbatched `vmap` (~86 GB in one
+> allocation); `autolens_workspace#450` set `batch_size=4` and the leg was
+> discharged by Release Integrate run 30901054267 on 2026-08-04
+> (`scripts/interferometer/start_here.py ... PASS (173.5s)`).
+>
+> Re-verified at retirement (2026-08-27):
+> - `autolens_workspace` and `autogalaxy_workspace` **both** carry `batch_size=4`
+> on the 48-start `MultiStartProdigy` in `scripts/interferometer/start_here.py`,
+> so the amendment's consequence 1 ("it is not autolens-specific") is closed too.
+> - The interferometer shard is green in the two most recent nightly Release
+> Integrate runs — 32804373015 (2026-08-25, `autogalaxy`) and 33073386315
+> (2026-08-27, `autolens`). No recurrence.
+> - The NUFFT/transformer hypothesis in "Notes for triage" was wrong: the
+> allocation was the multi-start batch, not a dense transformer matrix.
+>
+> The amendment's consequence 2 — `JAX_TRACEBACK_FILTERING=off` in the release
+> harness — was **not** done and is still open. It was split out as its own task:
+> `draft/bug/ci/jax_traceback_filtering_release_harness.md`.
+
+---
+
# interferometer/start_here.py OOM in nightly release-validation integrate leg
Filed: 2026-07-31 (backfilled from git)
diff --git a/complete/index.md b/complete/index.md
index 9b79da84..237a3b56 100644
--- a/complete/index.md
+++ b/complete/index.md
@@ -6,7 +6,7 @@ Token-light navigation over the finished-work records (schema:
only then grep a dated bucket. Curators: edit the band between the CURATED
markers; everything below GENERATED is rebuilt.
-1153 records across 7 buckets.
+1154 records across 7 buckets.
## Highlights
@@ -125,6 +125,7 @@ _(curate hard-won records here — survives regeneration.)_
- [jax-likelihood-mass-sensitivity](2026/08/jax-likelihood-mass-sensitivity.md) — Audit + fix for median-literal mass blindness across all jax_likelihood/datacube scripts (phase 2 of the findi…
- [jax-likelihood-smoke-pins-stale](2026/08/jax-likelihood-smoke-pins-stale.md)
- [jax-stall-shared-preloads-retime-refutation](2026/08/jax-stall-shared-preloads-retime-refutation.md)
+- [jax-traceback-filtering-release-harness](2026/08/jax-traceback-filtering-release-harness.md)
- [knowledge-board](2026/08/knowledge-board.md) — auto-closed on merge
- [lacosmic-cr-option-and-star-pass-decoupling](2026/08/lacosmic-cr-option-and-star-pass-decoupling.md) — both deliberately OPEN — closure is the default-flip decision, not this ship
- [latex-docstrings-invalid-escape-warnings](2026/08/latex-docstrings-invalid-escape-warnings.md)
diff --git a/dashboard.html b/dashboard.html
index 85f95f1a..ed2588ff 100644
--- a/dashboard.html
+++ b/dashboard.html
@@ -204,7 +204,7 @@
PyAutoMindDashboard
Intent. Priority. Flow.
Every task the Mind is holding. Tap a task's 📋 and its /start_dev command is on your clipboard — paste it into a Claude Code chat to route Claude straight to that task. Recent is the same work by date — what has been happening rather than what to do next.
-
1In flight
3Parked
5Planned
136Backlog
+
1In flight
3Parked
5Planned
135Backlog
Last updated 2026-08-27. This page is generated from active/, draft/ and the registry files, so it is only as current as they are. dashboard_refresh.yml re-renders it on every push to main — that heals a stale page, but not a stale prompt: a task that shipped without its prompt advancing to complete/ keeps rendering here as pickable backlog. Reconciling those is the refresh below.
docs — 12
diff --git a/dashboard.md b/dashboard.md
index cc89c9a0..26688199 100644
--- a/dashboard.md
+++ b/dashboard.md
@@ -45,7 +45,7 @@ anything you could not verify.
| [In flight](#in-flight) (`active/`) | 1 |
| [Parked](#parked) (`parked.md`) | 3 |
| [Planned](#planned) (`planned.md`) | 5 |
-| [Backlog](#backlog) (`draft/`) | 136 |
+| [Backlog](#backlog) (`draft/`) | 135 |
## Start here
@@ -247,7 +247,7 @@ Scoped but not started; some are not yet prompt files. Full detail in [`planned.
## Backlog
-**136** filed prompts, not started. Each section is sorted most-pickable first (priority, then size). **25** of them belong to an epic and are listed only under [Epics](#epics) below.
+**135** filed prompts, not started. Each section is sorted most-pickable first (priority, then size). **25** of them belong to an epic and are listed only under [Epics](#epics) below.
feature — 28
@@ -777,7 +777,7 @@ Scoped but not started; some are not yet prompt files. Full detail in [`planned.
-bug — 17
+bug — 16📋 LensCalc NumPy Hessian step is too coarse for multi-plane tracers — autogalaxy · large · supervised · high
@@ -907,14 +907,6 @@ Scoped but not started; some are not yet prompt files. Full detail in [`planned.
-📋 interferometer/start_here.py OOM in nightly release-validation integrate leg — autolens
-
-```
-/start_dev draft/bug/autolens/interferometer_release_leg_oom.md
-```
-
-
-
@@ -1522,12 +1514,11 @@ Continue the 'Expectation propagation (EP) campaign' epic. Its canonical state l
## Hygiene
-4 prompt(s) without a metadata header — they show no facets above. Re-home or re-run intake on them when touched.
+3 prompt(s) without a metadata header — they show no facets above. Re-home or re-run intake on them when touched.
Headerless prompts
-- `draft/bug/autolens/interferometer_release_leg_oom.md`
- `draft/docs/autolens_workspace/sampler_cli_output_workspace_sweep.md`
- `draft/docs/workspaces/plot_coverage_followups.md`
- `draft/research/autolens_profiling/cluster_gradient_search_benchmark.md`
diff --git a/draft/triage/nightly_release_blocked_eight_nights.md b/draft/triage/nightly_release_blocked_eight_nights.md
index 2d726090..075885d7 100644
--- a/draft/triage/nightly_release_blocked_eight_nights.md
+++ b/draft/triage/nightly_release_blocked_eight_nights.md
@@ -102,8 +102,8 @@ most of the set was already fixed or tracked by the time the streak was noticed.
| Night | Failing legs | Cause | Status |
|---|---|---|---|
| 07-28 | 3 × `FileNotFoundError: dataset/…/data.fits` (autogalaxy/multi, autolens/imaging) | auto-simulate guards pointed at the wrong simulator | **FIXED** — see below |
-| 07-30 | `interferometer/start_here.py` in **both** autolens and autogalaxy shards; `group/start_here.py` TIMEOUT | same OOM as 07-31, message eaten by the JAX traceback filter | `draft/bug/autolens/interferometer_release_leg_oom.md` |
-| 07-31 | `interferometer/start_here.py` → `RESOURCE_EXHAUSTED: Out of memory allocating 85898814480 bytes`; `delaunay.py` TIMEOUT | ~86 GB single allocation; XLA hang | as above; autolens_workspace_test#245 |
+| 07-30 | `interferometer/start_here.py` in **both** autolens and autogalaxy shards; `group/start_here.py` TIMEOUT | same OOM as 07-31, message eaten by the JAX traceback filter | **FIXED** — OOM `complete/2026/08/interferometer-start-here-integrate-oom.md`; filter `complete/2026/08/jax-traceback-filtering-release-harness.md` |
+| 07-31 | `interferometer/start_here.py` → `RESOURCE_EXHAUSTED: Out of memory allocating 85898814480 bytes`; `delaunay.py` TIMEOUT | 48-start `MultiStartProdigy` vmapped unbatched; XLA hang | **FIXED** — autolens_workspace#450 (`batch_size=4`), discharged by run 30901054267; autolens_workspace_test#245 |
| 08-01 | `guides/modeling/advanced/hierarchical.py` TIMEOUT; `delaunay.py` TIMEOUT | Nautilus deadlock; XLA hang | hierarchical **FIXED** (PyAutoFit#1443); #245 |
| 08-03 | `graphical/ep.py` — "`log_likelihood_function` is always returning `nan`" | EP initializer | `draft/bug/autofit/graphical_ep_nan_likelihood_release_leg.md` |
| 08-04 | `guides/results/database/start_here.py` `IndexError` | `samples_weight_threshold` pruning | **FIXED + MERGED** 08-04 09:40 — autolens_workspace#465, autogalaxy_workspace#202 |
@@ -171,8 +171,13 @@ this needs its own prompt.
- `draft/bug/autofit/covariance_interpolator_test_unseeded_rng.md` — Class B
root cause. Smallest and highest value: it alone unblocks a live release.
-- `draft/bug/autolens/interferometer_release_leg_oom.md` — filed 07-31, never
- issued; now updated with the 07-30 both-workspaces recurrence.
+- ~~`draft/bug/autolens/interferometer_release_leg_oom.md`~~ — **closed out
+ 2026-08-27.** Its OOM had already shipped on 07-31 (autolens_workspace#450,
+ `batch_size=4` on the 48-start `MultiStartProdigy`;
+ `complete/2026/08/interferometer-start-here-integrate-oom.md`), so the prompt
+ was retired to `complete/archive/shelved/`. The one live part of its 08-04
+ amendment — the JAX traceback filter that ate the 07-30 message — shipped as
+ PyAutoHeart#187 (`complete/2026/08/jax-traceback-filtering-release-harness.md`).
- `draft/bug/autofit/graphical_ep_nan_likelihood_release_leg.md` — new.
- autolens_workspace_test#245 (delaunay hang) — already tracked, no action here.
- Confirm on 08-05 that Check D self-healed and that PyAutoBrain#196 renders the