diff --git a/AGENTS.md b/AGENTS.md index efb3915a6..b318073a5 100644 --- a/AGENTS.md +++ b/AGENTS.md @@ -6,6 +6,12 @@ when the task needs them. Read [README.md](README.md) for the public overview. ## Where work lives +- Profiling campaign intent and pending tasks live in **PyAutoPulse** + (`campaigns.yaml`, `tasks/`, `CHECKIN.md`). Use its single-chat check-in first. + Mind holds bounded implementation phases and repository claims, linking the + canonical Pulse task. See `docs/pulse-task-migration.md`. + + - `draft///.md` holds unstarted prompts. Work types: `feature`, `bug`, `refactor`, `docs`, `test`, `release`, `maintenance`, `research`, and `triage` when genuinely unclear. `human_review` is reserved diff --git a/active.md b/active.md index 16d9d891a..356465e0f 100644 --- a/active.md +++ b/active.md @@ -1,5 +1,20 @@ # Active Tasks +## pulse-campaign-control-room +- issue: https://github.com/PyAutoLabs/PyAutoPulse/issues/6 +- issued: 2026-10-03 +- prompt: active/campaign_control_room.md +- session: Codex Work +- status: awaiting-merge +- worktree: /workspace/scratch/df4e4aa94b97 +- repos: + - PyAutoPulse: feature/campaign-control-room + - PyAutoMind: feature/pulse-task-migration +- summary: Single-chat campaign check-in and profiling task migration into Pulse. +- workspace-pr: https://github.com/PyAutoLabs/PyAutoPulse/pull/7 +- resume: Pulse#7 has 86 passing tests and scoped Heart GREEN. Merge Pulse before this Mind migration; then close out issue Pulse#6 and release both claims. No compute submitted. +- validation: Ruff, formatting, 86 tests, offline snapshot/feed check, 26 source hashes and copy-handler checks pass. Mind lifecycle check passes with one pre-existing batch calibration warning. + ## vis-lp-inspection-bundle - issue: https://github.com/PyAutoLabs/euclid_strong_lens_modeling_pipeline/issues/102 - issued: 2026-09-22 diff --git a/active/campaign_control_room.md b/active/campaign_control_room.md new file mode 100644 index 000000000..8418b49ab --- /dev/null +++ b/active/campaign_control_room.md @@ -0,0 +1,30 @@ +# Pulse campaign control room and task migration + +Type: feature +Target: PyAutoPulse +Repos: +- PyAutoPulse +- PyAutoMind +Status: awaiting-merge +Issue: https://github.com/PyAutoLabs/PyAutoPulse/issues/6 +Issued: 2026-10-03 +Filed: 2026-10-03 + +## Original request + +I have not looked at the PyAutoPulse dashboard yet but basically I want it to mirror PyAutoCortex was a table of active campaigns (profilijg specifci) a single copyable prompt at the top to get all updates which will mean all profiling work is managed in a single chat from now on. I also want all profiling tasks in PyAutoMind to move to PyAutoPulse as active tasks listed under the table. Then all the profiling info itself further down. For that initial prompt, it should be I can add to it if I want to direct a specific campaign suggest a specific idea albeit like Cortex I could just tell a prompt at the start. + +## Plan + +- Add a single editable check-in prompt above a campaign table. +- Give Pulse a validated campaign/task ledger and migrate profiling prompts with provenance. +- Preserve existing task gates and retain the detailed measurement dashboard below. +- Document one-chat check-ins, task updates and bounded development handoffs. +- Validate renderer, task links, migration coverage and repository checks. + +Tier: undeclared β€” merge mode: human /prm. + +Implementation: `PyAutoPulse/pulse/campaigns.py`, `campaigns.yaml`, `tasks/`, +`CHECKIN.md`, board/CLI integration and regression tests. Mind's migration record +and routing link replace duplicate queued work; historical completion records remain. +Project measurement producers and scientific acceptance criteria stay authoritative. diff --git a/dashboard.html b/dashboard.html index 32d0941aa..e97969d5a 100644 --- a/dashboard.html +++ b/dashboard.html @@ -231,7 +231,7 @@

PyAutoMindDashboard

Intent. Priority. Flow.

Every task the Mind is holding. Tap a task's πŸ“‹ and its start-dev skill prompt is on your clipboard β€” paste it into an AI assistant chat to route the assistant straight to that task. Recent is the same work by date β€” what has been happening rather than what to do next.

-
  • 2In flight
  • 2Human review
  • 6Parked
  • 8Planned
  • 258Backlog
+
  • 3In flight
  • 2Human review
  • 6Parked
  • 7Planned
  • 232Backlog

Last updated 2026-10-03. 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 β€” 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.

Fix KNN neighbor search for a partial final point blockπŸ› bugautoarraysmallsupervisedhigh

Forward-mode gradient is NaN for FitPositionsImagePairAll(Solved) β€” released default gradient_mode="forward"πŸ› bugautolenssmallsupervisedhigh

test_latent_euclid_variables_traces_under_jax_jit went red on main: total_source_flux differs by 6% between jax.jitβ€¦πŸ› bugeuclidsmallsupervisedhigh

@@ -275,7 +275,7 @@

Highest priority (filed as high) β€” showing 12 of 23

Local smoke env creation fails: smoke_install.sh flat pip chain vs the groupedβ€¦πŸ› bugpyautoheartmediumsupervisedhigh

Physical model check when speeding up smoke testsπŸ§ͺ testworkspacesmediumsupervisedhigh

Fix Sibson interpolation at internal edgesπŸ› bugautoarraylargesupervisedhigh

-

Fits a slot (ready to run unattended, cheapest to review first) β€” showing 12 of 125

+

Fits a slot (ready to run unattended, cheapest to review first) β€” showing 12 of 113

Guard every workspace script against untimed network downloadsπŸ§ͺ testautolens_workspacesmallsafemedium

Intake Agent silently drops unknown Type values and overrides declared Targetβ€¦πŸ› bugpyautobrainsmallsafemedium

worktree.sh bundle creation clobbers the unversioned root activate.shπŸ› bugpyautobrainsmallsafemedium

@@ -290,6 +290,7 @@

Fits a slot (ready to run unattended, cheapest to revie

Use readthedocs or migrate to GitHub docsπŸ”¬ researchpyautohandssmallsupervisednormal

In flight markdown version

Issued β€” each has an open GitHub issue and usually a branch.

+

Pulse campaign control room and task migration β€” issue #6 β€” issued 2026-10-03 β€” awaiting-merge β€” PRs: PyAutoPulse#7

Preserve the LensCalc evaluation field when the grid cap activates β€” issue #645 β€” issued 2026-10-02 β€” library-merged, workspace-release-gated β€” PRs: PyAutoGalaxy#646, autolens_workspace_test#343, autolens_profiling#365 β€” ⏳ pending release: PyAutoGalaxy

Support vis_lp-only products in inspection bundles β€” issued 2026-09-22

Pending release markdown version

@@ -392,20 +393,19 @@

Parked markdown version

-8 task(s) +7 task(s)

isothermal-ell-sph-oversampling-at-the-cusp β€” found 2026-08-09 β€” planned β€” NOT yet a prompt file; file one via /intake before starting

remote-mcp-deployment-tiers β€” issue #20 β€” filed 2026-07-21 β€” DESIGN-COMPLETE, build BLOCKED-ON-DEMAND β€” issue #20 holds the full auth/transport/hosting design + Richard/PyAutoMCP…

samples-parameter-paths β€” issue #1327 β€” filed 2026-07-08 β€” parked

piemass-potential β€” filed 2026-06-05 β€” planned

latent-nan-guard-honest-run β€” planned 2026-07-22

-

timing-noise-audit β€” issue #362 β€” filed 2026-10-02 β€” planned

knn-partial-point-block β€” issue #609 β€” filed 2026-10-02

sibson-internal-edge-continuity β€” issue #610 β€” filed 2026-10-02

Backlog markdown version

-

258 filed prompts, not started β€” sorted most-pickable first (priority, then size). 57 of them belong to an epic and are listed only under Epics below.

+

232 filed prompts, not started β€” sorted most-pickable first (priority, then size). 46 of them belong to an epic and are listed only under Epics below.

-bug β€” 79 +bug β€” 75

Fix KNN neighbor search for a partial final point blockπŸ› bugautoarraysmallsupervisedhigh

Forward-mode gradient is NaN for FitPositionsImagePairAll(Solved) β€” released default gradient_mode="forward"πŸ› bugautolenssmallsupervisedhigh

test_latent_euclid_variables_traces_under_jax_jit went red on main: total_source_flux differs by 6% between jax.jitβ€¦πŸ› bugeuclidsmallsupervisedhigh

@@ -416,7 +416,6 @@

Backlog

Sparse (w-tilde) inversion fits the unsubtracted image when a regular light profileβ€¦πŸ› bugautoarraymediumsupervisedhigh

Local smoke env creation fails: smoke_install.sh flat pip chain vs the groupedβ€¦πŸ› bugpyautoheartmediumsupervisedhigh

Fix Sibson interpolation at internal edgesπŸ› bugautoarraylargesupervisedhigh

-

Audit timing tests and profiling gates for measurement noiseπŸ› bugautolens_profilinglargesupervisedhigh

Fix release result/sample parameter-path regressionsπŸ› bughealth_fixestoo-largesupervisedhigh

EP test suite is not xdist-safe: tests share one on-disk output directoryπŸ› bugautofitsmallsafemedium

Interferometer galaxy_image_dict drops a mixed galaxy's ordinary light (dict merge overwrites)πŸ› bugautogalaxysmallsupervisedmedium

@@ -464,7 +463,6 @@

Backlog

multi_dataset/jax_likelihood/delaunay.py asserts the wrong likelihood on main (βˆ’72070 vs 1995)πŸ› bugautolens_workspace_testmediumsupervisednormal

PyAutoNerves emits a false "workspace is stale, git pull" warningβ€¦πŸ› bugautonervesmediumsupervisednormal

@PyAutoFit TransformedMessage.logpdf/pdf omit the transform JacobianπŸ› bugpriorsmediumsupervisednormal

-

jax_profiling/gradient/imaging/pixelization.py: 3.2% of its pin move is unattributedπŸ› bugworkspacesmediumsupervisednormal

autocti_workspace: the dataset_1d results/database example scripts have driftedπŸ› bugautocti_workspacelargesupervisednormal

MultiStartGradient searches ignore iterations_per_quick_update and live_visual_update while announcing the cadenceβ€¦πŸ› bugautofitlargesupervisednormal

Resolve release-profile timeout scripts deliberatelyπŸ› bughealth_fixestoo-largesupervisednormal

@@ -478,19 +476,16 @@

Backlog

Give the JAX PowerLaw deflections branch the same exact unit-vector transformβ€¦πŸ› bugautogalaxysmallsafelow

PowerLawMultipole with m=1 returns -inf / NaN deflections at slope exactly 2β€¦πŸ› bugautogalaxysmallsupervisedlow

Cortex conductor test resolves through the worktree symlink and fails in everyβ€¦πŸ› bugpyautobrainsmallsafelow

-

profile_lens_aggregator.py cannot run from the autolens_workspace_developer root: no config/ directoryπŸ› bugworkspacessmallsafelow

The reconstruction noise map is not the truncated posterior the NNLS solveβ€¦πŸ› bugautoarraymediumhuman-requiredlow

release-integrate: analyze-side foreign-path filter and a workspace-repo lint for tracked result filesπŸ› bugpyautoheartmediumsupervisedlow

Point-source JSON datasets record no resolution regimeπŸ› bugpyautolensmediumsupervisedlow

-

MGE likelihood_breakdown steps are cumulative and linear_gaussians is reported as 0πŸ› bugworkspacesmediumsafelow

Same-class q-clamp and bare-sqrt ellipticity gradient sites (follow-up to PyAutoGalaxy#631)πŸ› bugautogalaxy

euclid_dr1: every dataset rgb.jpg is upside down relative to the VIS FITSβ€¦πŸ› bugeuclidsupervised

-feature β€” 35 +feature β€” 32

Inversion.source_clumps_from needs a robust scale β€” thresholding against the raw maximum findsβ€¦βœ¨ featureautoarraysmallsupervisedmedium

autofit_assistant: a skill that owns loading the user's data and its selectionβ€¦βœ¨ featureautofit_assistantsmallsafemedium

-

Numba breakdown harness: perturb the instance so the operated-matrix memo cannot hideβ€¦βœ¨ featureautolens_profilingsmallsafemedium

Smoke profile cannot see a missing sampler dependency β€” add an importβ€¦βœ¨ featurepyautohandssmallsafemedium

Which other searches need prior-support handling β€” coverage audit after Prodigy✨ featureautofitmediumsafemedium

Give PyAutoFit searches a seed β€” today no search can be madeβ€¦βœ¨ featureautofitmediumsafemedium

@@ -509,7 +504,6 @@

Backlog

Skip the likelihood for assertion-violating samples on the JAX path✨ featureautofitlargesupervisednormal

Remote-MCP deployment tiers (2 + 3) for the results-inspector server✨ featureautofit_assistantlargehuman-requirednormal

Multi-plane time delays✨ featureautolenslargesupervisednormal

-

Search settings-estimation + profiling infrastructure (n_starts / batch_size / n_batch)✨ featureautolens_profilinglargesupervisednormal

Adopt oversampled PSFs in the start-here dataset chain (option a)✨ featureautolens_workspacelargesupervisednormal

Restore lensing-of-lens in the multiple_sources workspace example✨ featureworkspaceslargesupervisednormal

Follow-up to rectangular_adapt_cdf.md (issue #322) and Path A✨ featureautoarraytoo-largesupervisednormal

@@ -518,43 +512,13 @@

Backlog

Context: PyAutoLens issue #542 follow-up (Gap 1, deferred during the✨ featurejax_substructuretoo-largesupervisednormal

Context: PyAutoLens issue #542 follow-up (Gap 2, deferred during the✨ featurejax_substructuretoo-largesupervisednormal

dPIE: optional central-dispersion (sigma_0) parameterization✨ featureautogalaxysmallsupervisedlow

-

A gradient-cost probe: forward vs value_and_grad ms/eval and a strict FD checkβ€¦βœ¨ featureautolens_profilingsmallsafelow

Release board: local run_logs enrichment✨ featurepyautohandssmallsupervisedlow

Gallery runner: add visualization_upper + decide the modeling_visualization_jit tier✨ featureworkspacessmallsafelow

Deferred: make the *Power adapt classes the defaults (breaking)✨ featureautoarraylargehuman-requiredlow

Drop the 1e-3 prior centring for shear / multipole / ell_compsβ€¦βœ¨ featureautolens_inference

-research β€” 26 -

HPC campaign epic: run large modelling campaigns efficiently, token-cheaply and low-carbonπŸ”¬ researchautofitlargehuman-requiredhigh

-

Point-source A100 speed-up campaign: profile and optimize with the shared breakdownπŸ”¬ researchautolens_profilinglargesupervisedhigh

-

euclid_dr1: run the positions gate over DR1 and submit the remodel setπŸ”¬ researcheuclidlargehuman-requiredhigh

-

Deep research: Can we speed up Delaunay in PyAutoArray?πŸ”¬ researchautoarraytoo-largesupervisedhigh

-

Pair JAX/XLA env vars with measured compile and run times, per backendπŸ”¬ researchautonervestoo-largesupervisedhigh

-

Delaunay-family JAX modules never hit the persistent compilation cacheπŸ”¬ researchautoarraymediumsafemedium

-

Quick-update plotting cost β€” minutes per update, and it is not JAXβ€¦πŸ”¬ researchautolensmediumsafemedium

-

Linear-solver programme phase 3: GPU/vmap/A100 timing and parity rows for the solverβ€¦πŸ”¬ researchautoarraytoo-largesupervisedmedium

-

Use readthedocs or migrate to GitHub docsπŸ”¬ researchpyautohandssmallsupervisednormal

-

Re-baseline the slacs0008 acceptance parity after the HAP-dedupe fixπŸ”¬ researchpyautoreducesmallsafenormal

-

Kernel-CDF bandwidth defaults β€” config-dependent quality, investigate adaptivityπŸ”¬ researchautoarraymediumsafenormal

-

Is Intel macOS a supported platform, and what is the numpy-only contract?πŸ”¬ researchlibrariesmediumsafenormal

-

Checkerboard PSF-mismatch residual diagnostic β€” research + document + ingest papersπŸ”¬ researchpyautomemorymediumsafenormal

-

autofit_profiling: bootstrap the repo + general PyAutoFit profiling epicπŸ”¬ researchautofitlargesupervisednormal

-

Profile imaging pixelizations at production over-samplingπŸ”¬ researchautolens_profilinglargesupervisednormal

-

Assess remaining likelihood bottlenecks after certified solver integration on CPU and GPUsπŸ”¬ researchautolens_profilinglargesupervisednormal

-

Record harness-smoke runs for free Codex, a named OpenCode configuration and Geminiβ€¦πŸ”¬ researchautolens_assistantlow-mediumsupervisednormal

-

Sparse interferometer inversion: jax.grad of the log-evidence terms w.r.tπŸ”¬ researchautoarraysmallsupervisedlow

-

Should TransformedMessage carry its own support, rather than the prior?πŸ”¬ researchgraphical_epmediumhuman-requiredlow

-

Chase the ~6% flux scale between PyAutoReduce and legacy SLACS reductionsπŸ”¬ researchpyautoreducemediumsafelow

-

PyAutoArray Delaunay interpolator's pure_callback vs vmap β€” minor efficiency follow-upπŸ”¬ researchautoarraytoo-largesupervisedlow

-

HPC campaign epic β€” research report and phased planπŸ”¬ researchautofit

-

Energy and CO2e for PyAuto HPC campaigns: methods, inputs, worked estimateπŸ”¬ researchautofit

-

HPC campaign epic: what the codebase already hasπŸ”¬ researchautofit

-

euclid_dr1 HPC glue + Slurm mechanics β€” research reportπŸ”¬ researchautofit

-

Prior art: status, right-sizing and failure triage for large HPC fit campaignsπŸ”¬ researchautofit

-
-
-maintenance β€” 23 +maintenance β€” 22

Reduce the organism's per-session and per-task token load by half🧹 maintenanceorgansmediumsupervisedhigh

Workspace resolver fan-out: the hook, the smoke shims and the hardcoded paths🧹 maintenancepyautobrainlargesupervisedhigh

Keep caches off $HOME on HPC β€” phase 2 remainder: euclid pipelineβ€¦πŸ§Ή maintenanceworkspacessmallsupervisedmedium

@@ -562,7 +526,6 @@

Backlog

The workspace staleness warning keys off a value documented never to move🧹 maintenancehowtofitsmallsupervisednormal

Unregistered worktrees are invisible to the conflict guard🧹 maintenancepyautobrainsmallsupervisednormal

Mirror drifted library config keys into the workspace configs🧹 maintenanceworkspacessmallsafenormal

-

jax_compile/probe.py lost its cell builder with the searches tier β€” give profilingβ€¦πŸ§Ή maintenanceautolens_profilingmediumsupervisednormal

autolens_workspace_developer: broad stale-API rot (56 symbols, no CI)🧹 maintenanceautolens_workspace_developermediumsupervisednormal

wiki-currency's --check-version gate rots on every library main merge🧹 maintenancecimediumsupervisednormal

Dependency-cap refresh 2026-08: safe bumps, astropy 8 decision, two dead deps🧹 maintenancelibrariesmediumsupervisednormal

@@ -580,6 +543,28 @@

Backlog

Regenerate the 28 drifted .claude/hooks/session-start.sh copies after the organ-order change🧹 maintenancepyautomindsmall

+research β€” 19 +

HPC campaign epic: run large modelling campaigns efficiently, token-cheaply and low-carbonπŸ”¬ researchautofitlargehuman-requiredhigh

+

euclid_dr1: run the positions gate over DR1 and submit the remodel setπŸ”¬ researcheuclidlargehuman-requiredhigh

+

Deep research: Can we speed up Delaunay in PyAutoArray?πŸ”¬ researchautoarraytoo-largesupervisedhigh

+

Delaunay-family JAX modules never hit the persistent compilation cacheπŸ”¬ researchautoarraymediumsafemedium

+

Use readthedocs or migrate to GitHub docsπŸ”¬ researchpyautohandssmallsupervisednormal

+

Re-baseline the slacs0008 acceptance parity after the HAP-dedupe fixπŸ”¬ researchpyautoreducesmallsafenormal

+

Kernel-CDF bandwidth defaults β€” config-dependent quality, investigate adaptivityπŸ”¬ researchautoarraymediumsafenormal

+

Is Intel macOS a supported platform, and what is the numpy-only contract?πŸ”¬ researchlibrariesmediumsafenormal

+

Checkerboard PSF-mismatch residual diagnostic β€” research + document + ingest papersπŸ”¬ researchpyautomemorymediumsafenormal

+

Record harness-smoke runs for free Codex, a named OpenCode configuration and Geminiβ€¦πŸ”¬ researchautolens_assistantlow-mediumsupervisednormal

+

Sparse interferometer inversion: jax.grad of the log-evidence terms w.r.tπŸ”¬ researchautoarraysmallsupervisedlow

+

Should TransformedMessage carry its own support, rather than the prior?πŸ”¬ researchgraphical_epmediumhuman-requiredlow

+

Chase the ~6% flux scale between PyAutoReduce and legacy SLACS reductionsπŸ”¬ researchpyautoreducemediumsafelow

+

PyAutoArray Delaunay interpolator's pure_callback vs vmap β€” minor efficiency follow-upπŸ”¬ researchautoarraytoo-largesupervisedlow

+

HPC campaign epic β€” research report and phased planπŸ”¬ researchautofit

+

Energy and CO2e for PyAuto HPC campaigns: methods, inputs, worked estimateπŸ”¬ researchautofit

+

HPC campaign epic: what the codebase already hasπŸ”¬ researchautofit

+

euclid_dr1 HPC glue + Slurm mechanics β€” research reportπŸ”¬ researchautofit

+

Prior art: status, right-sizing and failure triage for large HPC fit campaignsπŸ”¬ researchautofit

+
+
docs β€” 16

Split lensing regimes: multi_galaxy / group / cluster (epic plan)πŸ“– docsautolenstoo-largesupervisedhigh

Advanced workspace guide: Preloads (PyAutoArray)πŸ“– docsworkspacestoo-largesupervisedhigh

@@ -642,47 +627,6 @@

Bundles Sets of INDEPENDENT tasks that make sense in one orchestrated session: an architect session plans them, subagents implement them, and every member still gets its own issue and its own PR β€” so the prm skill closes each one out unchanged. Not an epic: nothing here is ordered or phase-gated, and every member also appears in its usual section above β€” a bundle is an extra view of the backlog, never a replacement. Pinned bundles are the human record in bundles.md; auto bundles are recomputed from the backlog every time this page is rendered and are proposals, never records.

-point-source β€” bundle 2 β€” 3 task(s) Β· 8 pts Β· auto β€” proposed β€” ⚠️ theme(s) not in themes.md: euclid, jax-gpu -

Run this bundle β€” one session, one issue and one PR per member

- - - - - - - - - - - - - - - - - - - - - - - -
PromptRepoDifficultyPriorityStatus
Point-source A100 speed-up campaign: profile and optimize with the…autolens_profilinglargehighformalised
Point-source JSON datasets record no resolution regimepyautolensmediumlowformalised
Move the Witt–Wynne SIEP solver/projection into PyAutoLens once the…autolensmediumlowdraft
-
-
autoarray β€” bundle 1 β€” 4 task(s) Β· 7 pts Β· auto β€” proposed

Run this bundle β€” one session, one issue and one PR per member

- - - - - - - - - - - - - - - - - - - - - - - - - - -
PromptDifficultyPriorityStatus
Physical model check when speeding up smoke testsmediumhighformalised
rectangular.py parity contract: absolute-nats bound, and import…easymediumformalised
Sync remaining workspace config/priors copiessmallmediumformalised
MGE likelihood_breakdown steps are cumulative and linear_gaussians…mediumlowformalised
-
-
pyautoheart β€” bundle 1 β€” 4 task(s) Β· 6 pts Β· auto β€” proposed

Run this bundle β€” one session, one issue and one PR per member

+ + + + + + + + + + + + + + + + + + + + +
PromptDifficultyPriorityStatus
Physical model check when speeding up smoke testsmediumhighformalised
rectangular.py parity contract: absolute-nats bound, and import…easymediumformalised
Sync remaining workspace config/priors copiessmallmediumformalised
+
+
ci β€” 3 task(s) Β· 3 pts Β· auto β€” proposed β€” ⚠️ theme(s) not in themes.md: ci, robustness

Run this bundle β€” one session, one issue and one PR per member

+ + + + + + + + + + + + + + + + +
PromptRepoDifficultyPriorityStatus
SED chain waveband fit can spin forever in Nautilus exploration…euclidsmallhighformalised
euclid_dr1: every dataset rgb.jpg is upside down relative to the VIS…euclid---
+
+

Showing 8 of 39 auto bundles β€” pin one in bundles.md to keep it on the page.

Recent markdown version

The 50 newest things to happen to the work in hand, newest first β€” issued, parked, filed. Every other section on this page is laid out by state, which is exactly why none of them can answer β€œwhat has been happening?”. Shipped work is not here: it is read from complete/index.md, and a thousand records deep it would crowd out everything anyone can still act on. Showing the newest 10; … opens the next 10.

- - - - + + + + @@ -1046,18 +1016,12 @@

Bundles

- - - - - - - + @@ -1160,12 +1124,6 @@

Bundles

- - - - - - @@ -1252,12 +1210,6 @@

Bundles 2026-09-17

- - - - - - @@ -1297,6 +1249,24 @@

Bundles Inversion.source_clumps_from needs a robust scale β€” thresholding…

+ + + + + + + + + + + + + + + + + +
2026-10-02filedtiming-noise-audit2026-10-03issuedPulse campaign control room and task migration
2026-10-02
2026-10-02filedAudit timing tests and profiling gates for measurement noise
2026-10-01 filed Sparse interferometer terms: NUFFT ignores the mask origin and…
2026-10-01 filed Reconcile inference documentation with the current Cortex ledger…

Epics markdown version

@@ -1308,19 +1278,14 @@

Epics

PyAutoPulse β€” the cross-project dashboard layer above the <lib>_profiling project repos: instance registry, profiling-summary read contract, organ board, cockpit transition β€” ledger: complete/2026/10/pyautopulse-brain-board-cockpit.md β€” filed 2026-10-02; name decided 2026-10-02 (PyAutoPulse, key pulse); phase 1 COMPLETE 2026-10-02 (autolens_profiling#359 closed…

-Point-source (single-source) PointSolver CPU speed-up β€” 4 queued prompt(s), in order -

Point-source (single-source) PointSolver CPU speed-up β€” ledger: autolens_profiling/wiki/campaigns/point_source_image_plane_cpu.md (full record: results/notes/point_source_cpu_campaign.md) β€” phases 1-3 shipped (p2 + p3 released in 2026.9.26.1: PyAutoArray 7fa8d271, PyAutoLens 86054bbc); phase 4a (re-baseline +…

-

Runtime cells' A100 single_jit includes the post-compile warm-up β€” source-plane 0.642 msβ€¦πŸ› bugautolens_profilingsmallsupervisedlow

+Point-source (single-source) PointSolver CPU speed-up β€” 1 queued prompt(s), in order +

Point-source (single-source) PointSolver CPU speed-up β€” ledger: autolens_profiling/wiki/campaigns/point_source_image_plane_cpu.md (full record: results/notes/point_source_cpu_campaign.md) β€” phases 1-3 shipped (p2 + p3 released in 2026.9.26.1: PyAutoArray 7fa8d271, PyAutoLens 86054bbc); phase 4a (re-baseline +…

Set galaxy-scale PointSolver grid extents per workspace packageβ€¦βœ¨ featureautolens_workspacemediumsupervisednormal

-

Point-source source-plane chi-squared speed-up campaign β€” remaining candidates (phases 1–2e shipped)πŸ”¬ researchautolens_profilinglargesupervisedlow

-

Point-source (single-source) CPU campaign β€” carried leftovers and completion evidenceπŸ”¬ researchautolens_profilingsmallsupervisedlow

-Cluster PointSolver β€” robustness and performance β€” 3 queued prompt(s), in order -

Cluster PointSolver β€” robustness and performance β€” ledger: draft/research/autolens_profiling/cluster_pointsolver_speed.md β€” Expanded by human agreement 2026-10-01 to own robust analysis settings, numerical correctness and performance. Carries Source &…

+Cluster PointSolver β€” robustness and performance β€” 1 queued prompt(s), in order +

Cluster PointSolver β€” robustness and performance β€” ledger: https://github.com/PyAutoLabs/PyAutoPulse/blob/main/tasks/cluster_pointsolver_speed.md β€” Expanded by human agreement 2026-10-01 to own robust analysis settings, numerical correctness and performance. Carries Source &…

PointSolver error-behavior change: bisect the 2025-11β†’2026-05 candidates, then health-hardenπŸ› bugautolenstoo-largesupervisedhigh

-

Cluster PointSolver β€” robustness, analysis settings and performanceπŸ”¬ researchautolens_profilinglargesupervisednormal

-

PointSolver profiling cells: lensed quasar β†’ cluster runtime tier β†’ single/multi-source β†’β€¦πŸ”¬ researchautolens_profilinglargesupervisednormal

Expectation propagation (EP) campaign β€” 7 queued prompt(s), in order @@ -1353,7 +1318,7 @@

Epics

Model figures phase 6b2 β€” the model figure in SLaM pipeline stagesβ€¦βœ¨ featureworkspacesmediumsupervisedlow

autolens_inference β€” inference benchmarking repo, birth to first base run β€” ledger: autolens_inference/wiki/project/state.md

-

Linear-solver accuracy/tolerance programme β€” a standing autolens_profiling package for positive-only solver studies β€” ledger: autolens_profiling/wiki/campaigns/linear_solver_accuracy.md (contract: complete/2026/09/linear-solver-accuracy-study.md) β€” phase 2 SHIPPED 2026-09-30 β€” PyAutoArray#595 merged (7a89e19a0, forward polish; pending release) + autolens_profiling#357 merged…

+

Linear-solver accuracy/tolerance programme β€” a standing autolens_profiling package for positive-only solver studies β€” ledger: autolens_profiling/wiki/campaigns/linear_solver_accuracy.md (contract: complete/2026/09/linear-solver-accuracy-study.md) β€” phase 2 SHIPPED 2026-09-30 β€” PyAutoArray#595 merged (7a89e19a0, forward polish; pending release) + autolens_profiling#357 merged…

Cluster strong lensing β€” Source & Cluster arc β€” 12 queued prompt(s), in order

Cluster strong lensing β€” Source & Cluster arc β€” ledger: draft/feature/autolens/source_cluster_arc.md β€” In progress β€” 1a–1d and phase 3a audit merged (profiling#364, workspace_test#341). Phase 3b field-preserving cap fix open in…

@@ -1371,16 +1336,12 @@

Epics

Source & Cluster arc β€” magnification science, PointSolver trust, cluster extended sources✨ featureautolenstoo-largesupervisedhigh

-interferometer-likelihood-campaign β€” 9 queued prompt(s) β€” ⚠️ not in epics.md +interferometer-likelihood-campaign β€” 5 queued prompt(s) β€” ⚠️ not in epics.md

RAL PyAuto venv: third-party packages below the library floorsβ€¦πŸ› bugpyautohearteasysupervisedmedium

Reuse the fnnls Cholesky factor for the log det when edge-zeroed pixelsβ€¦βœ¨ featureautoarraysmallsupervisedlow

Interferometer likelihood campaign: chunk TransformerNUFFT.transform_mapping_matrix over columns and visibilities✨ featureautoarraymediumsupervisedmedium

Load-balance the prange direct_conv interferometer kernel (Delaunay stalls at 2 threads, autolens_profiling#332)✨ featureautoarraysmallsupervisedlow

Retune the interferometer numba gate 60 β†’ 70 (in-situ crossover, autolens_profiling#332)✨ featureautoarraysmallsupervisedlow

-

Interferometer decision matrix β€” fill the last cellβ€¦πŸ”¬ researchautolens_profilingsmallsupervisednormal

-

Interferometer fixed-mapper searches: reuse the W~ curvature matrix across likelihood callsβ€¦πŸ”¬ researchautolens_profilingmediumsupervisednormal

-

fnnls warm-start memo: stop it slowing scattered evaluation streams (interferometer CPU, autolens_profiling#332)πŸ”¬ researchautolens_profilingmediumsupervisedmedium

-

Interferometer W~ curvature matrix is FFT-bound on the mask extent: pruned paddedβ€¦πŸ”¬ researchautolens_profilingmediumsupervisednormal

certified-positive-solver β€” 2 queued prompt(s) β€” ⚠️ not in epics.md @@ -1399,14 +1360,6 @@

Epics

Mapping overlay follow-ups: forward regions= through the autogalaxy plot wrappers, fix degenerateβ€¦πŸ› bugautoarraymediumsafemedium

-
-mass-field β€” 1 queued prompt(s) β€” ⚠️ not in epics.md -

Remaining code consumers move to the flat fields= form: profiling, inference, JOSSβ€¦πŸ§Ή maintenanceautolens_profilinglargesupervisednormal

-
-
-streaming-visibilities β€” 1 queued prompt(s) β€” ⚠️ not in epics.md -

Campaign: interferometer streaming (array-free) vs in-memory β€” memory and time scalingβ€¦πŸ”¬ researchautolens_profilingsmallsupervisedhigh

-