[Jammy2211]

What changed

The graphical-ep campaign's science half gets its first Cortex science project.

• projects.yaml — new slope_hierarchy_scale row: status: active,
  assistant: autolens_assistant, local /mnt/c/Users/Jammy/Science/slope_hierarchy_scale,
  RAL /mnt/ral/jnightin/slope_hierarchy_scale, ledger wiki/project/state.md,
  witness results/**/*.json, remote: none (the private org repo is a human act).
• projects.yaml — slope_hierarchy retired via cortex.py retire (it answered its four
  N=5 goals and was wrapped up 2026-07-22), plus a hand edit of its local_path to the vault
  /mnt/c/Users/Jammy/Science/z_projects_complete/slope_hierarchy, so the row still says
  where the N=5 data lives.
• tasks/ — git mv slope_hierarchy/n25_scale_up.md slope_hierarchy_scale/;
  Project: retargeted; Gates: PyAutoFit#1405, PyAutoFit#1558, PyAutoFit#1560, PyAutoFit#1562;
  the stale "Ready when: graphical-ep phase 2 is issued" line replaced by the phase-2 gate
  evidence and a "Borrowed from" note; then cortex.py move … gated.
• dashboard.md / dashboard.html re-rendered from this branch's tree.

Outside this repo (the science tree, no PR surface): Science/slope_hierarchy was moved
intact — git history, remote, 17 commits — to Science/z_projects_complete/slope_hierarchy,
and Science/slope_hierarchy_scale was born with one commit (acce42a), scaffolded per
autolens_assistant/skills/start-new-project.md Phase 1 and borrowing
scripts/{ep,graphical,one_by_one,util}.py, simulators/sample.py, config/, activate.sh,
environment.yml and hpc/. Not borrowed: output/, dataset/, the old results/ and the
old wiki/project/ journal — the N=5 baseline is cited from the vault in the new
wiki/project/state.md, and the datasets are re-simulated at N=25.

Why

The human's ruling this session: phase 3 of graphical-ep runs as a fresh project, with
the wrapped-up N=5 tree vaulted and mined for code, rather than reopening a finished project.
Phase 3 was chosen over phase 4 because the ic50 checkout is dirty and has no results/
witness path.

Verification

• python3 scripts/cortex.py check — OK
• python3 -m pytest tests/ -q — 126 passed
• pyauto-brain cortex dashboard --cortex <worktree> --apply then --check — exit 0
• python3 scripts/cortex.py gates — the task's four refs resolve to PyAutoFit issue URLs
• new project: git log --oneline = one birth commit; no output/ or dataset/ content
  tracked; scripts/{util,one_by_one,graphical,ep}.py all import clean

The human's next acts, in order

1. (optional) gh repo create PyAutoLabs/slope_hierarchy_scale --private --source=/mnt/c/Users/Jammy/Science/slope_hierarchy_scale --push, then flip the row's remote:.
2. Merge this PR (projects.yaml is code — a human merge by design).
3. /cortex → python3 scripts/cortex.py move tasks/slope_hierarchy_scale/n25_scale_up.md ready once the four gate refs are judged cleared → ask to submit the N=25 runs.

Residue (surfaced, not acted on)

• tasks/slope_hierarchy/methods_writeup.md stays planned under the retired row — move it
  to slope_hierarchy_scale/ or drop it.
• The moment-matching cure decision behind PyAutoFit#1405 is still the human's.
• ic50_workspace (phase 4) untouched: its checkout is dirty and it has no witness path.

Closes #24

🤖 Generated with Claude Code

https://claude.ai/code/session_01JcyWRpcbRiAAYFqgFJarEr