Claude Code Plugins

Community-maintained marketplace

Feedback

figure-qa

@neuromechanist/research-skills
29
2

Use this skill to QA a scientific figure for journal compliance, alignment, palette correctness, and label legibility. Triggers on "QA this figure", "check this figure", "review my figure", "is this figure paper-ready", "validate figure", or when invoked proactively after a figure-generation skill produces output (unless the caller passes no-qa). Dispatches on input type (SVG, raster PNG/JPG/TIFF, Python plot script, or composed-figure directory) and runs the right programmatic checks plus a VLM rubric judgment pass.

Install Skill

Shared

Installs to .agents/skills, used by Codex, Amp, Warp, Cursor, OpenCode, and more.

CodexAmp
WarpCursorOpenCode
ClineGemini CLI
GitHub Copilot
Personal

Available across projects.

$npx skills-installer add @neuromechanist/research-skills/figure-qa --client shared
Project

Writes to .agents/skills.

$npx skills-installer add @neuromechanist/research-skills/figure-qa -p --client shared
Note: Review the skill instructions before using it.

SKILL.md

name figure-qa
description Use this skill to QA a scientific figure for journal compliance, alignment, palette correctness, and label legibility. Triggers on "QA this figure", "check this figure", "review my figure", "is this figure paper-ready", "validate figure", or when invoked proactively after a figure-generation skill produces output (unless the caller passes no-qa). Dispatches on input type (SVG, raster PNG/JPG/TIFF, Python plot script, or composed-figure directory) and runs the right programmatic checks plus a VLM rubric judgment pass.
version 0.1.0

Figure QA

Routes a scientific figure to an independent, fresh-context QA reviewer that detects the input type, runs the deterministic checks (fonts, palette, geometry, alpha, DPI) via helper scripts, and adds a VLM rubric judgment for the aesthetic dimensions. This skill is a thin dispatcher: the detection logic, exit-code contract, VLM rubric, and report format all live in references/figure-qa-procedure.md, and the deterministic engine lives in the figures plugin's agents/figure-qa-scripts/.

When to use

Activate when a figure needs a journal-submission QA pass, or proactively after a figure-generation skill (scientific-figure, svg-figure, svg-primitives, transparent-icons, ai-full-figure, plot-styling) produces output, unless the caller passes no-qa.

Why a fresh-context reviewer

A QA pass is more trustworthy from a reviewer that did not just author the figure. Run it in a separate context and pass only the figure path, input type (if known), and target journal. On tools without subagents, run the procedure inline.

Dispatch

In every branch the reviewer follows references/figure-qa-procedure.md (strict separation: scripts own ground-truth measurements, VLM owns aesthetic judgment).

  • Claude Code: Task(subagent_type: "figure-qa", ...) passing the figure path and target journal. Honor a no-qa opt-out by returning immediately.
  • Codex CLI: plugin installation exposes this skill, not a Codex subagent. To use a fresh-context Codex reviewer, first copy agents/templates/figure-qa.toml to ~/.codex/agents/ or .codex/agents/, then invoke that configured agent if the current Codex surface supports /agent. If no Codex subagent is configured or available, use the fallback branch.
  • Copilot CLI: plugin installation exposes this skill and, through .github/plugin/plugin.json, the .agent.md reviewer in agents/templates/. Invoke that configured agent when the current Copilot surface supports custom agents. If running outside a plugin install, copy agents/templates/figure-qa.agent.md to .github/agents/ or ~/.copilot/agents/. If no custom agent is available, use the fallback branch.
  • Fallback (no subagent support, or an interactive in-thread check): first locate the procedure ($CLAUDE_PLUGIN_ROOT/skills/figure-qa/references, else find . -type d -path '*/skills/figure-qa/references' | head -1); if it cannot be found, stop and tell the user to install the figures plugin rather than guessing checks. Then follow references/figure-qa-procedure.md directly.

The brain (do not duplicate into dispatch or agent shells)

  • references/figure-qa-procedure.md -- no-qa opt-out, script location, type detection, per-branch checks, exit-code contract, VLM rubric, and the exact report shape.
  • agents/figure-qa-scripts/check_{svg,raster,plot_script}.py (in the figures plugin) -- the deterministic engine, kept in place so other skills (e.g. svg-primitives) can call it directly.