mind: retire 13 prompts whose work already shipped - #296
Merged
Conversation
Picking a task off the dashboard and finding it already done has been a recurring waste. This is the sweep: every prompt under `draft/` audited by header, by fuzzy and content match against the 1104 completion records, and by checking the claimed work against the live tree where it is locally verifiable. Thirteen were spent. Each gets a dated record naming the evidence, with the original prompt folded in, and its cross-references repointed. Shipped, but the prompt never advanced: - `delaunay-nan-probe-ell-comps-validator` — autolens_workspace_test#277 (`995f0c9`); the shipping session set `Status: shipped` in place two days ago. - `remove-pynufft-legacy-transformer` — PyAutoArray#475, PyAutoGalaxy#583, PyAutoLens#709. - `history-policy-generated-block` — `policy/never_rewrite_history.md`, `repos_sync.py`'s history block and `check_history_blocks()` are all on main. - `interferometer-delaunay-nonpd-fitexception` — one defect filed twice from two repos; the autoarray-targeted twin advanced, this one did not (PyAutoFit#1408 + autolens_workspace#311, then PyAutoArray#436). - `unify-ai-assistant-workspace-readmes` — phase 2 shipped inside phase 1; `assistant-first-docs` names both workspace PRs. - `extra-galaxies-feature-parity` — its own final-phase record already said "ARC COMPLETE" and named this file as parent. - `python-312-minimum-census` — the census ran and the migration shipped in five phases; the prompt kept advertising the verbatim request it was executed from. - `dashboardify-brain-operational-surfaces` — the write-up exists and the Brain board it led to has been live and iterated on since 2026-08-23. - `profiling-agent-compile-axis-arc` + `jax-compile-time-profiling-absorbed` — all three children shipped 2026-08-10; both parents sat at `Priority: high`, top of the pick list, for two weeks. Superseded or retired in place, then left in `draft/`: - `multi-dataset-jax-likelihood-xla-stall` — folded into the jax-compile-stall epic 2026-08-23. - `latex-docstrings-invalid-escape-warnings` — the wider successor shipped 2026-08-20 across six repos. - `priors-and-messages-math-audit` — fully graduated 2026-08-18; open C-findings stay tracked under `bug/priors/`. Deliberately kept: partly-done trackers (`split_lensing_regimes`, `multi_galaxy_package`, `bug/priors/z_features`, `complete_archive_wiki`), prompts whose blocker merged but whose own work is outstanding (both `unpark_*_scaling_relation_slam`), and follow-up prompts that share a subject with a shipped record but not its scope (`plot_coverage_followups`, `notebook_setup_notebook_drift_siblings`, `cti_release_train_wiring`). Backlog 157 → 144. `lifecycle.py check`, `orphans` and `index --check` all OK. `dashboard.md` / `dashboard.html` are regenerated, so they also pick up the freshness banner added on the generator side (PyAutoBrain, same branch): the last-updated date, why a self-healing render is not a fresh backlog, and a one-tap 📋 copy of this whole reconciliation as a runnable instruction. Co-Authored-By: Claude <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01NcL3NFXVPW4xipE8wJfjSx
Conflicts were the three generated files plus one genuine collision. The collision: `delaunay_nan_probe_ell_comps_validator.md` was retired on both sides — by this sweep as `delaunay-nan-probe-ell-comps-validator`, and on main by the session that actually shipped the fix, as `delaunay-nan-probe-fix`. Theirs is kept and mine is dropped: it is the owning session's ship record, with the CI validation, the "the validator was right" finding and the probe's new shape. The sweep's version only pointed at the evidence from outside. The generated files (`complete/index.md`, `dashboard.md`, `dashboard.html`) were resolved by regenerating from the merged source of truth, not by hand: python3 scripts/lifecycle.py index --apply pyauto-brain intake --apply dashboard Rendered against PyAutoBrain main at 9063997 (#265 merged), so `intake dashboard --check` agrees with what CI will produce. Main had also retired `jax_point_source_point_smoke_sentinel.md` and `point_jax_vmap_parity_nondeterministic.md` — both merged cleanly, and both were prompts this sweep had inspected and deliberately left in the backlog. Backlog 144 -> 142. lifecycle check / index --check / orphans and registry_toc --check all OK. Co-Authored-By: Claude <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01NcL3NFXVPW4xipE8wJfjSx
github-actions
Bot
deleted the
claude/automind-prompts-dashboard-enl7mh
branch
August 25, 2026 14:01
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Ordering resolved — PyAutoLabs/PyAutoBrain#265 is merged (
9063997),mainwas merged in, and both checks are green.dashboard_refresh.ymlrenders this page with PyAutoBrain's default branch, which is why the generator change had to land first.Why
Picking a task off the dashboard and finding it already done has been a recurring waste. This is the sweep.
Every prompt under
draft/was audited three ways: by header, by fuzzy and content match against the completion records, and by checking the claimed work against the live tree where it is locally verifiable. Thirteen were spent. Each gets a dated record naming the evidence, with the original prompt folded in under## Original prompt, and its cross-references repointed.Shipped, but the prompt never advanced
remove-pynufft-legacy-transformerhistory-policy-generated-blockpolicy/never_rewrite_history.md,repos_sync.py's history block andcheck_history_blocks()are all onmaininterferometer-delaunay-nonpd-fitexceptionunify-ai-assistant-workspace-readmesassistant-first-docsnames both workspace PRsextra-galaxies-feature-paritypython-312-minimum-censusdashboardify-brain-operational-surfacesprofiling-agent-compile-axis-arc+jax-compile-time-profiling-absorbedPriority: high, top of the pick list, for two weeksSuperseded or retired in place, then left in
draft/multi-dataset-jax-likelihood-xla-stall— folded into thejax-compile-stallepic 2026-08-23.latex-docstrings-invalid-escape-warnings— the wider successor shipped 2026-08-20 across six repos.priors-and-messages-math-audit— fully graduated 2026-08-18; open C-findings stay tracked underbug/priors/.One collision, resolved in main's favour
delaunay_nan_probe_ell_comps_validator.mdwas retired on both sides — by this sweep, and onmainby the session that actually shipped the fix, asdelaunay-nan-probe-fix. Theirs is kept and mine is dropped: it is the owning session's ship record, carrying the CI validation, the "the validator was right" finding and the probe's new shape. The sweep's version only pointed at the evidence from outside. So twelve of the thirteen records here are new.Deliberately kept
Partly-done trackers (
split_lensing_regimes,multi_galaxy_package,bug/priors/z_features,complete_archive_wiki); prompts whose blocker merged but whose own work is outstanding (bothunpark_*_scaling_relation_slam); and follow-ups that share a subject with a shipped record but not its scope (plot_coverage_followups,notebook_setup_notebook_drift_siblings,cti_release_train_wiring).Two of those —
jax_point_source_point_smoke_sentinel.mdandpoint_jax_vmap_parity_nondeterministic.md— were retired onmainwhile this branch was open, and merged in cleanly.Validation
Backlog 157 → 142.
lifecycle.py check,orphans,index --checkandregistry_toc.py --checkall OK;intake dashboard --checkreports current against PyAutoBrainmain.The generated files were resolved by regenerating from the merged source of truth (
lifecycle.py index --apply,pyauto-brain intake --apply dashboard), never by hand. They also pick up the freshness banner from #265: the last-updated date, why a self-healing render is not a fresh backlog, and a one-tap 📋 copy of this whole reconciliation as a runnable instruction.