Vis Lens Figure Table

• Planning the presentation format for each result slot in a paper or report
• Deciding whether a leaderboard or ranking should be a table or a bar chart
• Evaluating whether ablation results communicate better as a matrix table or a figure
• Justifying borderline figure/table decisions before creating visualizations
• User invokes /autoskillit:vis-lens-figure-table

Critical Constraints

NEVER:

• Modify any source code files
• Do not litter the codebase with useless comments, TODO markers, or explanatory annotations — the skill output and diagram speak for themselves
• Create files outside {{AUTOSKILLIT_TEMP}}/vis-lens-figure-table/
• Force a figure when the reader needs exact numeric values for a leaderboard or lookup
• Force a table when the primary message is a trend, curve, or distributional shape

ALWAYS:

• Apply the decision rule below to every result slot in the experiment plan

• When the verdict is BOTH, recommend the table in the main paper and the figure in the appendix (or vice versa depending on the primary audience)

• Justify borderline decisions with explicit reference to the decision rule criteria

• BEFORE creating any diagram, LOAD the /autoskillit:mermaid skill using the Skill tool — this is MANDATORY

• If the Skill tool cannot be used (disable-model-invocation) or refuses this invocation, do NOT proceed with diagram creation. Abort this step and omit the diagram from output.

• Write output to {{AUTOSKILLIT_TEMP}}/vis-lens-figure-table/vis_spec_figure_table_{YYYY-MM-DD_HHMMSS}.md (relative to the current working directory)

• After writing the file, emit the structured output token as literal plain text with no markdown formatting on the token name (the adjudicator performs a regex match):

  diagram_path = /absolute/path/to/{{AUTOSKILLIT_TEMP}}/vis-lens-figure-table/vis_spec_figure_table_{...}.md
  

Analysis Workflow

Step 0: Parse optional arguments

If positional arg 1 (context_path) is provided and the file exists, read it to obtain IV/DV tables, H0/H1 hypotheses, controlled variables, and success criteria. If positional arg 2 (experiment_plan_path) is provided and exists, read the experiment plan for full methodology. Use this structured context as the foundation for Steps 1–4; skip the CWD exploration for these fields if the context file supplies them.

Step 1: Inventory Result Slots

Enumerate every metric, experiment result, or finding that needs to be communicated. For each result slot, classify it as one of:

• exact-value query: reader needs the precise number (leaderboard, lookup table)
• trend/curve: the message is a trajectory or change over time/steps
• distribution: the message is the shape of a distribution (histogram, violin, KDE)
• comparison/ranking: multiple systems compared by a scalar metric
• spatial/geometric: data has a spatial or structural layout (heatmap, scatter)
• ablation matrix: systematic removal/addition of components in a table grid

Step 2: Apply Decision Rule per Slot

For every BOTH verdict:

• Recommend the TABLE for the main paper body (exact values for citation)
• Recommend the FIGURE for the appendix (visual intuition for reviewers)
• Document the justification explicitly

Step 3: Emit yaml:figure-spec Blocks

For FIGURE and BOTH verdicts only (TABLE-only results produce no figure-spec). Then LOAD /autoskillit:mermaid and create a decision-tree diagram showing result-type → verdict flow for all result slots.

Output Template

# Decisional Layout Spec: {System / Experiment Name}

**Lens:** Decisional Layout (Decisional)
**Question:** Should this result be a figure or a table?
**Date:** {YYYY-MM-DD}
**Scope:** {What was analyzed}

## Layout Decision Summary

| Result Slot | Type | Verdict | Justification |
|-------------|------|---------|---------------|
| {main-results} | exact-value query | TABLE | Leaderboard — readers need precise numbers |
| {loss-curve} | trend/curve | FIGURE | Training trajectory is the message |
| {ablation} | ablation matrix | TABLE | Component grid; exact deltas matter |
| {distribution} | distribution | FIGURE | Shape is the message |
| {ranking-with-ci} | comparison/ranking | BOTH | Exact deltas + visual CI bands |

## Figure Specs

```yaml
# yaml:figure-spec — canonical schema (spec_version: "1.0")
figure_id: "fig-loss-curve"
figure_title: "Training Loss Trajectory"
spec_version: "1.0"
chart_type: "line"
chart_type_fallback: "scatter"
perceptual_justification: "Trend over steps is the message; table would obscure the trajectory shape."
data_source: "results/loss_curves.csv"
data_mapping:
  x: "global_step"
  y: "train_loss"
  color: "variant"
  size: ""
  facet: ""