Skip to content

Restructure PyAutoMind prompts by work type#23

Merged
Jammy2211 merged 1 commit into
mainfrom
claude/pyautomind-restructuring-prompt-2d77yy
Jun 26, 2026
Merged

Restructure PyAutoMind prompts by work type#23
Jammy2211 merged 1 commit into
mainfrom
claude/pyautomind-restructuring-prompt-2d77yy

Conversation

@Jammy2211

Copy link
Copy Markdown
Collaborator

Implements autoprompt/redesign.md. Reorganises the prompt repository so the first folder level describes the kind of work, and the second names the target repo/domain — so PyAutoBrain can route intent to the right specialist agent by reading the first path segment.

PyAutoMind organises intent by the kind of thinking required; PyAutoBrain uses that structure to choose the right reasoning agent.

New taxonomy

First folder = work type, second = target. Work types: feature/ bug/ refactor/ docs/ test/ release/ maintenance/ research/ experiment/, plus triage/ for prompts whose classification is still unclear.

Folder Future PyAutoBrain agent
feature/ feature planner
bug/ debugger
refactor/ refactor architect
docs/ documentation agent
test/ test engineer
release/ release engineer
maintenance/ hygiene agent
research/ research analyst
experiment/ prototype agent

Current active-prompt counts: feature 30, bug 21, research 8, docs 3, test 3, refactor 2, maintenance 2.

What moved

  • All 69 active prompts moved into <work-type>/<target>/ via git mv (history preserved). The old repo-named top-level folders (autoarray/, autofit/, autogalaxy/, autolens/, autolens_assistant/, autobrain/feature/pyautobrain/, autobuild/, workspaces/, visualization/, …) are gone.
  • Numbered topic series (jax_substructure, weak, cluster, priors) kept together under their dominant work type rather than scattered across folders, preserving ordering and their trackers.
  • One ambiguous prompt (jax_zero_contour.md) parked in triage/ with a manual-review note.

Examples of moved prompts:

  • autoarray/oversampling.mdfeature/autoarray/oversampling.md
  • autogalaxy/spherical_bug.mdbug/autogalaxy/spherical_bug.md
  • autofit/latent_class_redesign.mdrefactor/autofit/latent_class_redesign.md
  • autoarray/delaunay_research.mdresearch/autoarray/delaunay_research.md
  • autolens_assistant/science_project_collaborator_clone.mddocs/autolens_assistant/...
  • priors/* (audit series) → bug/priors/*

Preserved as lifecycle / meta (not work-types)

issued/, z_features/, z_vault/, shelved/ keep their own names; autoprompt/ remains the meta folder for prompts about this repo's own infrastructure.

References updated

  • planned.md prompt paths; the z_features trackers (cluster / jax_substructure / weak) and the priors in-folder tracker.
  • Repaired some pre-existing rotten links to already-shipped cluster/weak prompts (repointed to issued/).
  • scripts/status.sh inventory is now work-type aware; AGENTS.md layout and the start_dev / create_issue / register_and_iterate skill docs updated.
  • README: new "Prompt taxonomy" section (work-type → agent map, good prompt paths, optional metadata-header convention, migration note). Added ROUTING.md and triage/README.md.

Compatibility notes

Routing keys off @RepoName references in a prompt's body, not its folder — so the skills accept both old <target>/<name>.md and new <work-type>/<target>/<name>.md paths during the transition. PyAutoBrain agents are not implemented here; only the taxonomy + metadata are defined, per the prompt.

Validation

  • bash scripts/status.sh runs clean and lists the new work-type inventory.
  • A link-resolution sweep across the z_features trackers and the priors series passes (all relative .md links resolve).
  • No live references to removed folders remain outside intentional historical records (issued/, complete.md, queue.md DONE entries).

Follow-up tasks for PyAutoBrain agent routing

  • Implement the agents in PyAutoBrain (PyAutoAgent) that consume the first-folder work type per ROUTING.md.
  • Re-home triage/jax_zero_contour.md once its work type is confirmed (likely test/autolens_workspace_test/).
  • Classifications are subjective; a few (e.g. the whole priors audit under bug/, nufft_simulator_chunking under feature/ vs bug/) are one git mv to adjust if preferred.

🤖 Generated with Claude Code

https://claude.ai/code/session_01MMS6B2CqwaLwBfWGZRtCT9


Generated by Claude Code

Implements autoprompt/redesign.md. The first folder level now describes the
KIND of work (feature/bug/refactor/docs/test/release/maintenance/research/
experiment, plus triage/ for unclear cases); the second names the target repo
or domain. This replaces the old repo-named top-level folders (autoarray/,
autofit/, autogalaxy/, autolens/, …) so PyAutoBrain can route intent to the
right specialist agent by reading the first path segment.

- Moved all 69 active prompts into <work-type>/<target>/ via git mv (history
  preserved). Numbered topic series (jax_substructure, weak, cluster, priors)
  kept together under their dominant work type. One unclear prompt parked in
  triage/ with a manual-review note.
- Lifecycle/meta folders (issued/, z_features/, z_vault/, shelved/, autoprompt/)
  are NOT work-types and keep their own names.
- Updated references so paths resolve: planned.md, the z_features trackers
  (cluster/jax_substructure/weak), and the priors in-folder tracker. Also
  repaired pre-existing rotten links to shipped cluster/weak prompts (→ issued/).
- Updated scripts/status.sh inventory, AGENTS.md layout, and skill docs
  (start_dev z_features audit list + examples, create_issue, register_and_iterate)
  to the work-type layout; routing still keys off @reponame in the body, so old
  <target>/<name>.md paths continue to resolve during transition.
- README: new "Prompt taxonomy" section (work-type → PyAutoBrain agent map,
  good prompt paths, optional metadata-header convention, migration note);
  added ROUTING.md and triage/README.md.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01MMS6B2CqwaLwBfWGZRtCT9
Copilot AI review requested due to automatic review settings June 26, 2026 18:19

Copilot AI left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Pull request overview

This PR restructures PyAutoMind prompts into a new taxonomy where the first path segment indicates work type (feature/bug/refactor/docs/test/release/maintenance/research/experiment), enabling PyAutoBrain to route work based on folder classification. It also updates core documentation, skills, and trackers to reflect the new layout and introduces ROUTING.md + triage/ guidance for uncategorized prompts.

Changes:

  • Define and document the work-type → agent routing contract (ROUTING.md) and update repository docs (README.md, AGENTS.md) accordingly.
  • Update operational tooling and skills to understand the new path layout (scripts/status.sh, skills/*), plus refresh trackers/registries to point at moved prompts.
  • Add new prompts organized under the new taxonomy (including new triage/ holding area and new work-type/target prompt files).

Reviewed changes

Copilot reviewed 14 out of 82 changed files in this pull request and generated no comments.

Show a summary per file
File Description
z_features/weak_shear.md Updates weak-lensing epic tracker links to new/issued prompt locations.
z_features/jax_substructure_simulator.md Updates tracker links to feature/jax_substructure/*.
z_features/cluster_lensing.md Updates cluster epic tracker links to issued/ and feature/cluster/*.
triage/README.md Adds documentation for the non-routed triage holding area.
triage/jax_zero_contour.md Adds a triaged prompt with an HTML comment indicating it needs manual classification.
test/workspaces/smoke_workspace_fixes.md Adds a test-oriented prompt capturing smoke-test failures and follow-up fixes.
test/workspaces/restore_workspace_test_likelihood_baselines.md Adds a test-oriented prompt to restore absolute NumPy regression baselines in _workspace_test repos.
test/autolens_workspace_developer/mge_jit_regression_rebaseline.md Adds a test/developer prompt to re-baseline a profiling regression constant.
skills/start_dev/start_dev.md Updates /start_dev documentation and parsing rules to accept work-type paths.
skills/register_and_iterate/register_and_iterate.md Updates queue/registry prompt-path documentation for work-type layout.
skills/create_issue/SKILL.md Updates /create-issue usage examples for work-type paths.
scripts/status.sh Makes inventory reporting work-type aware (plus separate lifecycle/meta section).
ROUTING.md New routing contract documenting work-type → PyAutoBrain agent mapping.
research/autolens_workspace_developer/jax_jit_profiling.md Adds a research prompt about profiling JIT compilation latency.
research/autobuild/git_docs.md Adds a research prompt about GitHub/ReadTheDocs documentation strategy.
refactor/autogalaxy/einstein_radius_jit_native_seed_finder.md Adds a refactor prompt describing a JAX-native seed-finder approach.
README.md Updates repo layout + adds “Prompt taxonomy” documentation and migration note.
planned.md Updates planned registry entries to new prompt paths.
maintenance/workspaces/read_through_issues.md Adds a maintenance prompt with workspace cleanup/tasks list.
maintenance/autolens_profiling/polish.md Adds a maintenance prompt for consolidating and validating profiling workflows/results.
feature/workspaces/restore_multiple_sources_lensing_of_lens.md Adds a workspace feature prompt gated on an external PyAutoLens issue.
feature/workspaces/group_subhalo_sensitivity.md Adds a workspace feature prompt for group subhalo sensitivity mapping.
feature/workspaces/group_los_halos.md Adds a workspace feature prompt for group LOS-halo examples.
feature/weak/5_likelihood_function.md Adds a feature prompt to write weak-lensing likelihood tutorial script.
feature/weak/4_modeling.md Adds a feature prompt to implement weak-lensing modeling modules mirroring imaging API.
feature/pyautobrain/health.md Adds a feature prompt to implement a PyAutoBrain Health Agent.
feature/pyautobrain/feature.md Adds a feature prompt describing a PyAutoBrain Feature Agent design/behavior.
feature/pyautobrain/build.md Adds a feature prompt describing a PyAutoBrain Build Agent design/behavior.
feature/jax_substructure/6_deflection_equivalence_test.md Adds a feature prompt for a new deflection-equivalence integration test.
feature/jax_substructure/5_prng_key_vmap_noise.md Adds a feature prompt to thread PRNG keys through Poisson-noise APIs for vmap.
feature/jax_substructure/4_vmap_batched_simulation.md Adds a feature prompt for vmapped batched simulation and memory considerations.
feature/jax_substructure/3_simulator_jax_e2e.md Adds a feature prompt wiring scan/vmap ray-tracing through PSF+noise for E2E JIT simulate.
feature/jax_substructure/2_tracer_lax_scan.md Adds a feature prompt replacing nested loops with jax.lax.scan for multi-plane tracing.
feature/jax_substructure/1_vmap_subhalo_deflections.md Adds a feature prompt for vmapped per-plane deflection computation and tests.
feature/cluster/5_profiling.md Adds a feature prompt to add cluster likelihood profiling scripts.
feature/autolens_assistant/script_to_notebook.md Adds a feature prompt for narrative-docstring script → notebook conversion tooling.
feature/autolens_assistant/science_project_per_project_literature.md Adds a feature prompt defining project-local vs shared literature handling.
feature/autolens_assistant/science_project_assistant_ref.md Adds a feature prompt for assistant resolution/provenance pinning in science projects.
feature/autolens_assistant/science_project_api_gate.md Adds a feature prompt for reusing the assistant’s API validation gate in projects.
feature/autolens_assistant/portable_user_defaults.md Adds a feature prompt about environment discovery and optional portable defaults.
feature/autolens_assistant/live_visual_update_context.md Adds a feature prompt to surface live visualization options in assistant guidance.
feature/autogalaxy/point_source_light.md Adds a feature prompt proposing a point-source light profile and constraints.
feature/autogalaxy/piemass_potential.md Adds a feature prompt to implement missing PIEMass lensing potential.
feature/autofit/ep_lbfgs_jax.md Adds a feature prompt about scaling EP framework / alternative optimizers.
feature/autoarray/rectangular_multi_submesh.md Adds a feature prompt for multi-sub-mesh rectangular adaptation (Path B).
feature/autoarray/rectangular_adapt_constant_split_guard.md Adds a feature prompt to guard invalid mesh+regularization combinations.
feature/autoarray/oversampling.md Adds a feature prompt to implement oversampled PSF convolution support.
feature/autoarray/nufft_simulator_chunking.md Adds a feature prompt to add chunking to NUFFT simulator path to avoid OOMs.
feature/autoarray/multiwavelength_inversion.md Adds an (incomplete/draft) feature prompt for multiwavelength inversion approach.
feature/autoarray/arcsecond_to_decimal.md Adds a feature prompt to change arcsecond tick label formatting via plotting API.
docs/workspaces/preloads_advanced_workspace_guide.md Adds a docs prompt for an advanced Preloads guide in workspaces.
docs/workspaces/add_vincken_2026_wiki_and_cite_in_euclid.md Adds a docs prompt to ingest/cite a paper in downstream paper repos.
docs/autolens_assistant/science_project_collaborator_clone.md Adds a docs prompt to clarify collaborator cloning + assistant refer-back model.
bug/priors/z_features.md Updates priors/messages audit tracker references for new repo/taxonomy.
bug/priors/13_collapse_prior_and_message.md Adds a bug-tracker prompt for a large refactor/design discussion.
bug/priors/12_single_source_density_refactor.md Adds a bug-tracker prompt for density single-source-of-truth refactor design.
bug/priors/11_transformed_message_semantics_doc.md Adds a bug-tracker prompt documenting transform semantics and potential rename.
bug/priors/10_fixed_message_cache_growth.md Adds a bug-tracker prompt about unbounded cache growth in FixedMessage.
bug/priors/09_prior_property_tests.md Adds a bug-tracker prompt proposing property-based tests for priors/messages.
bug/priors/08_relative_width_modifier_safety.md Adds a bug-tracker prompt about sigma sign/zero safety in width modifiers.
bug/priors/07_log_prior_normalisation_convention.md Adds a bug-tracker prompt about inconsistent log-prior normalization conventions.
bug/priors/06_normal_message_sigma_negative_unchecked.md Adds a bug-tracker prompt about NormalMessage sigma validation and JAX behavior.
bug/priors/05_inv_beta_suffstats_clamp_noop.md Adds a bug-tracker prompt about a clamp no-op in beta sufficient-stat inversion.
bug/priors/04_truncated_normal_log_partition_incomplete.md Adds a bug-tracker prompt about truncated normal log-partition missing Gaussian term.
bug/priors/03_gamma_from_mode_wrong_formula.md Adds a bug-tracker prompt about incorrect GammaMessage.from_mode formula.
bug/priors/02_uniform_logpdf_array_handling.md Adds a bug-tracker prompt about UniformPrior.logpdf scalar-only boundary handling.
bug/priors/01_log_gaussian_with_limits_crash.md Adds a bug-tracker prompt about LogGaussianPrior.with_limits crash.
bug/autolens/pixelization_eager_vs_jit_divergence.md Adds a bug prompt investigating eager vs JIT divergence in pixelization likelihoods.
bug/autogalaxy/spherical_bug.md Adds a bug prompt flagging a likely spherical/geometry naming issue.
bug/autogalaxy/nfw_truncated_potential_accuracy.md Adds a bug prompt about NFWTruncated potential self-consistency failure.
bug/autofit/messages_xp_stack_jax_trace.md Adds a bug prompt about JAX tracing failures from xp.array([...]) patterns.
bug/autofit/factor_graph_3_14_instance_iteration.md Adds a bug prompt about FactorGraphModel instance shape regression on Python 3.14.
bug/autoarray/rect_adapt.md Adds a bug prompt about potential plotting alignment issues for rectangular meshes.
AGENTS.md Updates AI-agent guidance to the new prompt taxonomy and lifecycle/meta folders.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

@Jammy2211 Jammy2211 merged commit ef7eca0 into main Jun 26, 2026
1 check passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants