Paper Plan

If none exist, ask for a short description of the paper's contribution and evidence.

Workflow

Step 1: Extract Claims and Evidence

Read all available narrative and result documents and extract:

• core claims
• evidence for each claim
• known weaknesses
• recommended framing

Build a claims-evidence matrix:

| Claim | Evidence | Status | Section |
|-------|----------|--------|---------|
| [claim 1] | [exp A, metric B] | Supported | 3.2 |
| [claim 2] | [exp C] | Partially supported | 4.1 |

Step 2: Choose Paper Structure

Select a section structure that fits the contribution and page budget.

Standard IEEE Trans control paper structure:

I. Introduction
II. Preliminaries and Problem Formulation
III. Main Results (Controller/Observer Design + Stability Analysis)
IV. Simulation Results
V. Conclusion
Appendix: Proofs

Variations:

• Split Section III for complex contributions (e.g., III. Observer Design → IV. Controller Design → V. Stability Analysis)
• Merge Preliminaries into Introduction if brief
• Add Related Work as I-A or separate Section II if literature is dense

Step 3: Section-by-Section Planning

For each section, specify:

• the section goal
• which claims belong there
• which figures or tables are referenced there
• the expected length
• the citations required

For multi-agent formation control papers, ensure the method and experiment sections explicitly cover:

• agent dynamics and assumptions
• communication topology assumptions
• disturbance / delay / obstacle assumptions if relevant
• scalability with agent count
• what is theorem-backed vs simulation-backed vs engineering-only

Step 4: Detect Required Figure Families

Before writing the figure table, explicitly decide whether the paper needs:

• data figures from results
• non-data illustrations such as architecture, control loop, algorithm flow, communication topology, or scenario figures

For multi-agent formation control papers, check these non-data figures explicitly:

• Hero Architecture: overall pipeline or controller architecture
• Scenario Figure: agents, formation geometry, obstacles, targets, workspace, or communication links
• Algorithm Flow / Control Loop: when the method logic is not obvious from equations alone
• Communication Topology Figure: when graph structure is central to the contribution

Do not leave these implicit.

Step 5: Figure Plan

Write a figure plan in this exact shape:

## Figure Plan

| ID | Type | Render Module | Description | Data Source | Priority |
|----|------|---------------|-------------|-------------|----------|
| Fig 1 | Hero Architecture | paper-illustration | Overall control pipeline and baseline comparison focus | derived from method and setup | HIGH |
| Fig 2 | Scenario Figure | paper-illustration | Agent formation, topology, targets, and obstacles | derived from setup | HIGH |
| Fig 3 | Line Plot | paper-figure | Tracking error vs. time | figures/tracking.json | HIGH |
| Fig 4 | Bar Chart | paper-figure | Ablation results | figures/ablation.json | MEDIUM |
| Table 1 | Comparison Table | paper-figure | Main results vs. baselines | figures/main_results.json | HIGH |

Routing rules:

• Line Plot, Bar Chart, Grouped Bar, Scatter Plot, Heatmap, Box Plot, Violin Plot, Multi-panel, and Comparison Table -> paper-figure
• Hero Architecture, System Architecture, Algorithm Flow, Control Loop, Communication Topology, Scenario Figure, and Method Illustration -> paper-illustration
• use manual only when the user already has the asset or the figure is intentionally hand-drawn

This table is consumed downstream by /paper-writing for automatic module dispatch.

Step 6: Citation Scaffolding

For each section, list the required citation families and anchor references.

Citation rules:

• never fabricate BibTeX from memory
• verify author, year, and venue
• prefer published versions over preprints when available
• mark uncertain items with [VERIFY]

Step 7: Outline Review

Send the outline to gpt-5.4 for review with emphasis on:

• logical flow
• claim-evidence alignment
• missing experiments or analysis
• page-budget feasibility
• figure-plan completeness

For multi-agent formation control topics, ask the reviewer to check:

• whether assumptions are explicit enough for IEEE reviewers
• whether non-data figures are missing
• whether robustness and scalability evidence is sufficient

Step 8: Output

Write PAPER_PLAN.md with this structure:

# Paper Plan

## Metadata
- Title:
- Venue:
- Type:
- Page budget:
- Section count:

## Claims-Evidence Matrix

## Section Plan

## Figure Plan

## Citation Plan

## Risks Before Writing

Quality Gates

Before finishing, verify:

• every major claim has some evidence assignment
• unsupported claims are marked honestly
• the figure plan includes all required non-data figures
• each figure has a Render Module
• the plan can be consumed directly by /paper-writing