Figure capture TODO — for the agent in the history_pages worktree

Hand-off list for the agent with MCP access to the local Galaxy (worktree /Users/jxc755/projects/worktrees/galaxy/branch/history_pages, MCP galaxy-notebooks). Everything below should be obtainable from the existing UC1/UC2/UC3 histories, notebooks, and extracted workflows — no new analysis.

Already captured (in vault/papers/galaxy-notebooks/figures/): uc1_notebook_full, uc1_fig1_heatmap, uc3_volcano_embed, uc3_volcano_as_pdf_directive, uc3_pca_embed, uc3_pca_as_pdf_directive. (The uc2/uc3 notebook fulls were generated but not kept — regenerate if a notebook montage is wanted.)

Known UC1 IDs (from usecases/UC1_PAPER_INTEGRATION.md): notebook page eafb646da3b7aac5; extracted workflow 33b43b4e7093c91f; heatmaps 29e36fb8642bf5ed (have) / 579ae69ccbd17e45 (need); history 48916fac0de9a85d. UC2/UC3 IDs: pull from usecases/UC2_DEBRIEF*.md / UC3_DEBRIEF*.md, or list_pages / list_workflows via MCP.

For every capture: record the source object ID in a one-line manifest beside the PNG, use the light theme, a consistent window width, and 2× (retina) resolution.

STATUS (capture pass complete): all P1 done; P2 revision panel done; report continuity + quality re-captures documented but not done. Details/caveats in FIGURE_CAPTURE_REPORT.md; source IDs in figures/MANIFEST.md.

P1 — completes the three referenced manuscript figures (highest value)

• Figure 1, graph half — uc1_fig1_workflow_graph.png (workflow 33b43b4e7093c91f, all 14 nodes).
• Figure 2, second heatmap — uc1_fig1_heatmap2.png (+_full), dataset 579ae69ccbd17e45, matched to uc1_fig1_heatmap.png.
• Figure 3a — ATAC before/after — uc3_fig3a_workflow_degenerate.png (2 inputs/0 tools, extracted from the off-graph notebook’s two re-uploaded PNG references) + uc3_fig3a_workflow_5step.png. Same zoom.
• Figure 3b — ChIP caller/comparator — uc2_fig3b_caller_5step.png + uc2_fig3b_comparator_29step.png + optional uc2_single_condition-pinned_34step.png.

P2 — strengthening captures (optional, cheap from these histories)

• Report continuity — NOT done; needs an invocation render. Procedure in FIGURE_CAPTURE_REPORT.md.
• Revision provenance / agent authorship — uc1_revision_panel.png (page b887d74393f85b6d, 4 agent revisions badged (AI)). CAVEAT: shows agent authorship but not the user/agent/restore mix nor the GATA1-null correction — no real UC page carries all three edit_source kinds (see report).
• PDF renderer figure — skipped; the four existing UC3 PDF shots already cover it.

Quality re-captures (worth redoing the notebook fulls)

• UC1 done — uc1_notebook_full.png re-captured with history 48916fac0de9a85d set active (GET history/set_as_current?id=… inside the browser session, then /pages/editor?id=eafb646da3b7aac5); right panel now shows the populated MRSA history. [ ] UC2/UC3 still pending — same procedure (UC2 history 96d9e11f37f34b29 / page 42a2c611109e5ed3; UC3 history 241d84796a24640a / page a7e42332dab8f5db); their notebook fulls are also currently absent from figures/, so regenerate.
• Both full-app + content-cropped shots per notebook — see report.

Not capturable here (assembly/drawing, noted for completeness)

• Figure 3 panel composition (dashed seam arrow, before/after layout) — assembled from the P1 graph captures.
• Data-model and authoring-modes diagrams — drawn, not captured (see figures.md optional list).