Skip to content

sidiangongyuan/codex-skills-library / skills/paper-framework-figure-studio-pro

paper-framework-figure-studio-pro

Verified

Use when planning a source-grounded method overview, architecture, pipeline, system/data-flow, mechanism, or agent-workflow figure for a CS or deep-learning paper. Produces an editable figure brief before image generation or manual redrawing; not for plots, tables, or general prose.

Paper Framework Figure Studio Pro

This is a lightweight local adapter inspired by c-narcissus/paper-framework-figure-studio-pro. It keeps the useful paper-grounded S0-S5 co-design flow while avoiding the upstream package's heavy asset vendoring, mandatory checkpoint zip behavior, and fixed personal response rules.

Default Boundaries

  • Work in conversation by default. Do not create project files, reports, checkpoints, generated images, or restore bundles unless the user explicitly asks for them.
  • Treat image generation as opt-in. S2 and S5 are planning handoff stages unless the user explicitly asks to generate images.
  • Keep the workflow human-in-the-loop. Execute at most one public stage per response unless the user explicitly requests a compact text-only bundle such as "do S0/S1 together."
  • Stop after S5. The human chooses the final candidate, performs manual redrawing/editing, and decides final caption and paper placement.
  • Route final plot/table polish, Matplotlib/LaTeX edits, captions, and rendered validation to $paper-visual-craft.

Input Discipline

Before designing the figure, identify the available evidence:

  • Paper source: PDF, LaTeX, abstract/method text, notes, or user summary.
  • Target role: method overview, architecture, pipeline, data flow, mechanism, or agent workflow.
  • Venue/style constraints: CVPR/ICCV/ECCV/NeurIPS/ICLR/ICML/CoRL, double column, grayscale safety, line-art preference, space budget.
  • Existing visuals: reference figures, current draft figure, or explicit instruction not to inspect an existing paper diagram.
  • User constraints: must-show modules, forbidden claims, terminology, symbols, and whether image generation is allowed.

If inputs are incomplete, state the assumptions and ask only for missing facts that materially affect the diagram contract. Do not invent paper modules, symbols, arrows, metrics, or evidence.

Local S0-S5 Workflow

Use references/workflow.md when a task needs the full staged process.

  • S0 Paper foundation: inventory sources, extract core modules, terms, equations/symbols, data/control/model flow, evidence anchors, and ambiguity risks.
  • S1 Figure strategy: define the reader question, figure type, story path, semantic graph, visible text budget, style family, and source-grounded prompt contract.
  • S2 Sketch exploration: only if explicitly requested, generate or describe first-round candidate directions. Default to text-only candidate briefs.
  • S3 Direction select: compare candidates against the source contract and user preference; record what to keep, avoid, simplify, or repair.
  • S4 Formal candidate brief: prepare the final candidate matrix or image prompts with strict source-fidelity, symbol, connector, and density rules.
  • S5 Candidate image: only if explicitly requested, generate final raster candidates. Otherwise hand off prompt packages or a manual-redraw brief.

For ordinary requests, start with S0/S1 and return a concise figure brief rather than a full staged dossier.

Source-Faithful Figure Contract

Every visible entity, label, symbol, connector, and visual metaphor must be:

  • directly supported by the paper/user material, or
  • explicitly marked as an inference with the reasoning chain, or
  • marked unsupported and excluded from the figure.

Use these checks before proposing prompts or candidates:

  • Do arrows have evidence for both source and target semantics?
  • Are variables, metrics, weights, losses, or thresholds shown as edge/port/tag information instead of peer modules when appropriate?
  • Does one symbol/color/line style mean exactly one thing?
  • Is the diagram explaining the paper's actual contribution rather than a generic architecture pattern?
  • Can dense details move to caption, legend, or prose without weakening the first-glance story?

If the contract is not supportable, stop at the risk register and ask for the missing paper detail or user decision.

Default Output

Unless the user requested files or images, answer in conversation with:

  • figure role
  • reader question
  • source-backed core modules
  • semantic flow
  • visible text contract
  • style / surface recommendation
  • must-include / must-exclude
  • risks or unsupported parts
  • candidate directions
  • next action

Use templates/figure-brief-template.md when the user asks for a structured brief or when a worker needs a handoff.

Reference Routing

  • Read references/source-map.md before discussing provenance, upstream licensing, or vault updates.
  • Read references/workflow.md before running a full S0-S5 planning session.
  • Use templates/figure-brief-template.md for reusable handoff briefs.
writing-skillsobra/superpowersUse when creating new skills, editing existing skills, or verifying skills work before deploymentStars: 322kUpdated Claude CodeCodexVerifiedInstall
xlsxanthropics/skillsUse this skill any time a spreadsheet file is the primary input or output. This means any task where the user wants to: open, read, edit, or fix an existing .xlsx, .xlsm, .xltx, .csv, or .tsv file (e.g., adding columns, computing formulas, formatting, charting, cleaning messy data); create a new spreadsheet from scratch or from other data sources; or convert between tabular file formats. Trigger especially when the user references a spreadsheet file by name or path — even casually (like "the xlsx in my downloads") — and wants something done to it or produced from it. Also trigger for cleaning or restructuring messy tabular data files (malformed rows, misplaced headers, junk data) into proper spreadsheets. The deliverable must be a spreadsheet file. Do NOT trigger when the primary deliverable is a Word document, HTML report, standalone Python script, database pipeline, or Google Sheets API integration, even if tabular data is involved.Stars: 198kUpdated Claude CodeCodexVerifiedInstall
shipping-and-launchaddyosmani/agent-skillsPrepares production launches. Use when preparing to deploy to production, or when asking what needs to be in place before shipping. Use when you need a pre-launch checklist, when setting up monitoring, when planning a staged rollout, or when you need a rollback strategy.Stars: 119kUpdated Claude CodeCodexVerifiedInstall
childhoodthedotmack/claude-memBabysit a shipped release after merge, publication, or deployment. Verify published packages, production behavior, installed plugins, and late CI or user reports until the observation window is clean; use babysit for PR readiness before release.Stars: 118kUpdated Claude CodeCodexVerifiedInstall