academic-figure

Create, revise, or audit academic data/result figures for CS/AI/ML papers. Data/result plots default to Python-generated editable SVG with CS/AI/ML-specific design rules for benchmarks, ablations, tra

SKILL.md

GitHub

Academic Figure

CS/AI/ML academic-figure router. 实验数据图默认交付 Python/matplotlib 可编辑 SVG,并执行 CS/AI/ML 图表设计 gate。模型框架图、架构图、overview 图和复杂机制图不属于本 skill 的自动绘制范围;如论文需要此类图,只能输出人工绘制需求、证据清单、caption 草案或 figure blueprint notes。

Router Protocol

  1. Read manifest.yaml. It declares always_load files, axes, and references.on_demand.
  2. Read every file listed under always_load. These are the skill's binding rules — not reference material.
  3. Apply the loaded material as constraints:
    • stance.md defines Python-only plotting, figure contract, visual policy, and scope.
    • red-lines.md defines absolute prohibitions. Do not negotiate these.
    • output-contract.md defines deliverables per mode.
    • anti-patterns.md defines known failure modes and their correct alternatives.
  4. Select exactly one mode from the manifest. If ambiguous, ask one concise clarification only when data source or target use is missing. Requests for model framework, architecture, overview, or mechanism diagrams use figure-blueprint only to produce manual_figure_needed notes; never render images, SVG, or prompts.
  5. Echo the selected mode to the user before executing.
  6. Reach for references/ only when the manifest's references.on_demand condition is satisfied.

Modes

Mode Use when
chart-from-data Data files or numeric results, needs publication plot with CS/AI/ML chart design gate
figure-blueprint Wants figure suggestions for a paper section
figure-audit Existing figure reviewed for publication readiness
figure-revision Existing figure needs revision

Agent Dispatch

agents/figure_agent.md is dispatched by the orchestrator at Step 6.4. The agent returns figure artifacts, scripts, SVG paths, and reports; it must not independently edit project source code or experimental data.

Completion Criteria

  • chart-from-data: Figure Contract, CS/AI/ML chart design gate, Python script, source data, editable SVG, QA report — all pass.
  • figure-audit: Every QA item has pass/fail status and concrete remediation.
  • figure-blueprint: Every suggested figure maps to a paper claim and data/evidence source.
  • figure-revision: Revised artifact or instructions, QA report, unchanged evidence traceability.