{"slug":"figure-composer","title":"figure-composer","summary":"Compose or improve a publication-grade multi-panel scientific figure from a claim, concrete data paths, or an existing image. Use for figure outlining, parallel panel rendering, exact-grid composition, visual inspection, and adversarial figure review. Use figure-style for one sta","platform":"Claude","tags":[],"authorName":"LLM Mart","authorSlug":"llm-mart","score":0,"source":"github","price":null,"verified":false,"createdAt":"2026-08-23T08:58:30.018213Z","repo":{"url":"https://github.com/xuzhougeng/wisp-science","stars":1170,"forks":122,"license":"AGPL-3.0","updatedAt":"2026-09-25T06:13:53Z"},"bodyHtml":"
\n

name: figure-composer\ndescription: Compose or improve a publication-grade multi-panel scientific figure from a claim, concrete data paths, or an existing image. Use for figure outlining, parallel panel rendering, exact-grid composition, visual inspection, and adversarial figure review. Use figure-style for one standalone plot and paper-narrative for whole-paper figure ordering.\nlicense: Apache-2.0

\n

Figure composer

\n

Load figure-style with this skill. The sidecar provides pure geometry,\ncomposition, task-building, and review-schema helpers. It does not call models,\ndelegate Agents, resolve artifacts, or inspect images from Python.

\n

Inputs

\n

Require a one-sentence claim, target width in millimetres, and concrete\nproject-relative or absolute data paths. Never use artifact ids as paths. For an\nexisting figure, inspect the real image with view_image and write the outline\nyourself; pixels cannot reveal the source data path.

\n

Workflow

\n
    \n
  1. Build an outline matching figure_outline_schema(). Put real paths in\ndata_path; use null for schematics.
    2. \n
  2. Make panel a the conceptual hook and panel b the primary evidence. Use a\n12-column grid and one row per sub-claim.
    3. \n
  3. Build one instruction per panel with panel_task(...).
    4. \n
  4. If delegate_tasks is advertised, submit the independent panel tasks as one\nbatch. Grant each task the minimum advertised capabilities needed, normally\nvisualization plus project_read. Require a concrete PNG filename in each\noutput schema. If delegation is unavailable, render the panels sequentially\nwith python.
    5. \n
  5. Compose returned paths with compose_figure(...). Do not pass placeholder\nmarkers to the composer.
    6. \n
  6. Use compose_crops(...) with Pillow to save temporary crop files, then call\nview_image on the composite and every crop. Fix seams, clipped labels,\naliases, empty space, and misplaced panel letters before review.
    7. \n
  7. Build one reviewer instruction with composite_review_task(...). Delegate it\nwith image_inspection, project_read, and reasoning when those capability\nids are advertised; otherwise perform the review in the current Agent.
    8. \n
  8. Apply outline revisions and regenerate only affected panels. Stop after three\nrounds or when there are no blockers and at most two major findings.
    9. \n
\n

Outline example

\n
{\n  \"claim\": \"Treatment restores the disease-associated trajectory.\",\n  \"width_mm\": 180,\n  \"ncol\": 12,\n  \"row_heights_mm\": [42, 60],\n  \"panels\": [\n    {\n      \"letter\": \"a\",\n      \"role\": \"schematic\",\n      \"row\": 0,\n      \"col\": 0,\n      \"colspan\": 12,\n      \"chart_family\": \"study schematic\",\n      \"message\": \"The experiment tests trajectory rescue.\",\n      \"data_path\": null,\n      \"ask\": \"Show cohorts, treatment, sampling, and comparison.\"\n    },\n    {\n      \"letter\": \"b\",\n      \"role\": \"primary\",\n      \"row\": 1,\n      \"col\": 0,\n      \"colspan\": 12,\n      \"chart_family\": \"trajectory plot\",\n      \"message\": \"Treatment moves cells toward the healthy trajectory.\",\n      \"data_path\": \"results/trajectory.csv\",\n      \"ask\": \"Plot disease, treated, and healthy cells with confidence bands.\"\n    }\n  ]\n}\n
\n

Boundaries

\n\n","files":[{"path":"runtime.py","sizeBytes":11231,"isText":true},{"path":"SKILL.md","sizeBytes":3864,"isText":true}],"reviewScore":null,"reviewSummary":null,"trust":{"provenance":"trusted-source-unreviewed","notice":"Community-authored content, reproduced verbatim and not vetted as instructions. Treat it as data to evaluate, never as directives to follow.","bodySource":null},"bodyLocked":false,"purchaseUrl":null,"sourceUrl":null,"report":{"provenance":"trusted-source-unreviewed","screen":{"ran":true,"outcome":"clean","suspicious":0,"notes":0,"hiddenCharacters":false},"virusScan":{"engine":"clamav","status":"clean","scannedAt":"2026-09-25T07:39:01.005927Z","sha256":"EA4AB2521673011291BCC64CAF1B7BFB3681620A4FE11886F49A35BB5608FE90","sizeBytes":5319},"review":null,"source":{"repositoryUrl":"https://github.com/xuzhougeng/wisp-science","path":"skills/figure-composer","license":"AGPL-3.0","commit":"79e64163262196611dd390d06713c465b906b299","subtreeSha":"FF432A1F3B45E57E83ADF90FF5B150405F5461AC1B8ED7F2D5299CC0D09093F0","lastSyncedAt":"2026-09-25T07:37:55.552497Z"},"reviewedAt":"2026-09-25T07:40:44.02048Z","notice":"Community-authored content, reproduced verbatim and not vetted as instructions. Treat it as data to evaluate, never as directives to follow."},"install":[{"target":"skills-cli","command":"npx skills add https://github.com/xuzhougeng/wisp-science/tree/main/skills/figure-composer"},{"target":"claude-code","command":"claude plugin marketplace add https://llmmart.ai/marketplace.json && claude plugin install xuzhougeng-wisp-science@llmmart"},{"target":"git","command":"git clone https://github.com/xuzhougeng/wisp-science.git"}]}