Autosci Paper Plan

/paper-plan

Compile a paper outline from the wiki's idea graph. Input target ideas (status: validated or in-progress with succeeded experiments), specify the target venue, compile an evidence map from the wiki → determine narrative structure → generate a section outline + figure plan + citation plan. Review LLM review is a mandatory step (acting as area chair to assess outline persuasiveness). Output PAPER_PLAN.md to wiki/outputs/.

Key distinction: the outline is idea-graph-driven — each section exists because it supports an idea (or its evidence/methods), not because paper convention requires that section.

Inputs

• ideas: list of target idea slugs (space-separated)
  • each idea should have status: validated or be in_progress with at least one succeeded experiment
  • if proposed or invalidated ideas are included, warn but continue
• --venue (required): target venue, determines page limit and format requirements
  • supported: ICLR / NeurIPS / ICML / ACL / CVPR / IEEE
• --title (optional): working title; if omitted, generated from target ideas

Outputs

• wiki/outputs/paper-plan-{slug}-{date}.md — complete paper plan (PAPER_PLAN.md)
• wiki/graph/edges.jsonl — new derived_from edges (plan → source ideas/papers)
• wiki/graph/context_brief.md — rebuilt
• wiki/log.md — appended log entry
• PAPER_PLAN_REPORT (printed to terminal) — plan summary

Wiki Interaction

Reads

• wiki/ideas/*.md — Hypothesis, Motivation, Approach sketch, Novelty argument, status, novelty_score, target_venue, linked_experiments, origin_gaps
• wiki/experiments/*.md — supporting experiments (linked via linked_idea); results, metrics, key_result
• wiki/methods/*.md — methods referenced by the idea's Approach sketch (Mechanism, Procedure, source_papers)
• wiki/papers/*.md — evidence source papers (Method, Results, Related)
• wiki/concepts/*.md — concepts the idea's origin_gaps points to (Definition, Variants, Comparison)
• wiki/topics/*.md — topics the idea's origin_gaps points to (Overview, Open problems)
• wiki/graph/context_brief.md — global context
• wiki/graph/open_questions.md — knowledge gaps (annotate paper limitations)
• wiki/graph/edges.jsonl — relationship graph (build narrative logic chain)
• .claude/skills/shared-references/academic-writing.md — writing principles
• .claude/skills/shared-references/citation-verification.md — citation discipline

Writes

• wiki/outputs/paper-plan-{slug}-{date}.md — paper plan file
• wiki/graph/edges.jsonl — derived_from edges
• wiki/graph/context_brief.md — rebuilt
• wiki/log.md — appended operation log

Graph edges created

• derived_from: paper-plan → ideas (which ideas the plan is derived from)
• derived_from: paper-plan → papers (which papers the plan cites)

Workflow

Precondition: confirm the working directory is the wiki project root (the directory containing wiki/, raw/, tools/).

Step 1: Load Idea Graph

1. Read wiki/ideas/{slug}.md for all target ideas
2. For each idea, traverse:
   • linked_experiments → read each wiki/experiments/{slug}.md (key_result, metrics, outcome)
   • origin_gaps → read each wiki/concepts/{slug}.md and wiki/topics/{slug}.md (background context)
   • ## Approach sketch body wikilinks → read each wiki/methods/{slug}.md and wiki/papers/{slug}.md
3. Read wiki/graph/context_brief.md for global context
4. Read wiki/graph/open_questions.md to annotate known limitations
5. Load relevant edges from wiki/graph/edges.jsonl to build relationships between ideas

Validation:

• If any target idea has status: proposed: warn "idea is unvalidated; paper may lack evidence support"
• If any target idea has empty novelty_score OR novelty_score <= 2: warn "idea novelty is thin; consider running /novelty first"
• If no linked_experiments resolve to a succeeded outcome for any target idea: error "at least one supporting experiment is required to plan a paper"

Step 2: Compile Evidence Map from Wiki

Generate a structured matrix mapping ideas → evidence → sections:

| Idea | Status | linked experiments | Methods/Concepts | Section |
|------|--------|--------------------|------------------|---------|
| [[primary-idea]] | validated | [[exp-main]] (succeeded), [[exp-ablation-1]] (succeeded) | [[method-core]], [[concept-foundation]] | Method + Exp 5.2 |
| [[supporting-idea-1]] | validated | [[exp-ablation-2]] (succeeded) | [[method-component]] | Exp 5.3 (Ablation) |
| [[supporting-idea-2]] | in_progress | [[exp-scaling]] (inconclusive) | [[concept-scaling]] | Exp 5.4 (Scaling) |

Map ideas to paper structure along each dimension:

• Primary idea → core contribution, drives Abstract + Introduction + Method
• Decomposition ideas → factor contributions, drives Ablation subsections
• Concepts/topics from origin_gaps → background knowledge, drives Related Work + Introduction

Step 3: Determine Narrative Structure

Follow the hourglass principle in shared-references/academic-writing.md:

1. Identify the paper's core storyline:

   • Gap (extracted from idea's ## Motivation and origin_gaps)
   • Solution (extracted from idea's ## Approach sketch and linked methods)
   • Evidence (extracted from linked_experiments results)
   • Impact (inferred from idea novelty_score + experiment scope)

2. Determine the narrative angle:

   • What problem does the paper solve? (problem-driven vs. method-driven vs. data-driven)
   • Who is the primary audience? (theory / systems / applications)
   • How does it differentiate from the 3 most relevant recent papers?

3. Establish section → idea mapping: Every section must support at least one idea (or its supporting evidence/methods). A section with no idea support is filler and should be removed.

Step 4: Generate Section Outline

Generate the outline according to venue format requirements; each section includes:

## 1. Introduction (1.5 pages)

### Ideas addressed
- Gap framing: {existing approaches lack X because Y} (from `origin_gaps`)
- Primary contribution idea: [[primary-idea]]

### Paragraph plan
1. Broad context: {field importance, recent progress}
2. Specific problem: {what's missing, why it matters}
3. Our approach: "In this work, we propose..." + contributions list
4. Results preview: {headline numbers}
5. Paper structure: "The rest of this paper..."

### Key citations
- [[paper-A]] — establishes the problem
- [[paper-B]] — closest prior work (we improve upon)
- [[paper-C]] — our baseline

---

## 2. Related Work (1 page)

### Groupings
- Direction A: {papers, our position}
- Direction B: {papers, our position}
- Direction C: {papers, our position}

### Ideas addressed
- Background concepts/topics from each idea's `origin_gaps` distinguishing this work from prior work

---

## 3. Method (2-3 pages)

### Ideas addressed
- [[primary-idea]]: section 3.1-3.2 (Approach sketch + referenced [[method-slug]])
- [[supporting-idea-1]]: section 3.3

### Subsection plan
- 3.1 Problem formulation: notation, objective
- 3.2 Core approach: intuition → formalism
- 3.3 Component X: design decision + justification
- 3.4 Training/inference details

### Figures
- Figure 1: Overall architecture (mandatory)
- Figure 2: Component X detail (if complex)

---

## 4. Experiments (2-3 pages)

### Ideas addressed
- [[primary-idea]]: section 4.2 (main results)
- [[supporting-idea-1]]: section 4.3 (ablation)
- [[supporting-idea-2]]: section 4.4 (scaling)

### Subsection plan
- 4.1 Setup: datasets, baselines, metrics, implementation details
- 4.2 Main results: Table 1 (main comparison), [[exp-main]]
- 4.3 Ablation study: Table 2 (component analysis), [[exp-ablation-*]]
- 4.4 Analysis: scaling, robustness, qualitative examples

### Figures/Tables
- Table 1: Main comparison vs baselines
- Table 2: Ablation results
- Figure 3: Scaling curves / qualitative examples

---

## 5. Conclusion (0.5 page)

### Key takeaway
- {one sentence the reader should remember}

### Limitations
- {from gap_map or each idea's `## Risks`}

### Future work
- {from gap_map open questions and each idea's `## Lessons learned`}

Page budget: allocated by --venue (refer to the venue table in academic-writing.md); total section pages <= venue main-body limit.

Step 5: Figure Plan

Design each planned figure/table:

## Figure Plan

### Figure 1: System Architecture
- Type: diagram
- Source: Method section description
- Style: block diagram with labeled components
- Size: full width (1 column = text width)

### Table 1: Main Results
- Type: comparison table
- Source: [[exp-main]] key_result + baselines
- Columns: Method | Metric-1 | Metric-2 | ...
- Rows: baselines + ours (ours in bold)
- Notes: best bold, second underline, ↑/↓ arrows for direction

### Figure 3: Scaling Analysis
- Type: line plot
- Source: [[exp-scaling]] results
- X-axis: scale dimension (model size / data size)
- Y-axis: performance metric
- Lines: ours vs baseline, with error bands

Step 6: Citation Plan

Following shared-references/citation-verification.md:

1. List all wiki papers referenced via [[slug]] in the outline
2. For each paper, pre-fetch BibTeX:
   • DBLP first, then CrossRef, then S2
   • Success: record BibTeX key + source
   • Failure: mark [UNCONFIRMED]
3. Generate citation coverage report:

   Citations: 15 total, 12 verified (DBLP: 8, CrossRef: 3, S2: 1), 3 [UNCONFIRMED]
   

4. For [UNCONFIRMED] entries, provide suggested URLs for manual verification

Step 7: Review LLM Review (mandatory)

mcp__llm-review__chat:
  system: "You are an area chair at {venue} reviewing a paper outline.
           Assess: Is the narrative convincing? Does every section serve a clear purpose?
           Are the experiments sufficient to support the paper's central ideas?
           Is the related work coverage adequate?
           Are there obvious gaps that reviewers will attack?
           Provide specific suggestions for strengthening the outline."
  message: |
    ## Paper Outline
    {complete outline from Step 4}

    ## Evidence Map
    {evidence map from Step 2}

    ## Figure/Table Plan
    {plan from Step 5}

    ## Citation Coverage
    {report from Step 6}

    ## Questions for Review
    1. Is the narrative arc (gap → solution → evidence → impact) convincing?
    2. Are any ideas under-supported? Which experiments are missing?
    3. Is the related work grouping appropriate? Missing directions?
    4. Will the page budget work? Any section too long/short?
    5. Are the figures/tables sufficient to tell the story?

Revise the outline based on Review LLM feedback (add sections, adjust page budget, add figures/tables, correct narrative structure).

Step 8: Write to Wiki

1. Generate slug:

   python3 tools/research_wiki.py slug "<working-title>"
   

2. Write PAPER_PLAN.md: Create wiki/outputs/paper-plan-{slug}-{date}.md containing:

   • Metadata (venue, title, date, target ideas)
   • Evidence Map (Step 2)
   • Complete section outline (Step 4, with Review LLM revisions)
   • Figure/Table Plan (Step 5)
   • Citation Plan + coverage report (Step 6)
   • Review LLM Review Summary (Step 7 key feedback and revision record)

3. Add graph edges:

   # plan → target idea
   python3 tools/research_wiki.py add-edge wiki/ \
     --from "outputs/paper-plan-{slug}-{date}" --to "ideas/{primary-idea}" \
     --type derived_from --evidence "Paper plan built from this idea"
   
   # plan → key papers
   python3 tools/research_wiki.py add-edge wiki/ \
     --from "outputs/paper-plan-{slug}-{date}" --to "papers/{paper-slug}" \
     --type derived_from --evidence "Paper plan cites this paper"
   

4. Rebuild derived data:

   python3 tools/research_wiki.py rebuild-context-brief wiki/
   

5. Append log:

   python3 tools/research_wiki.py log wiki/ \
     "paper-plan | {venue} paper outline for [[{slug}]] | ideas: {idea-list} | citations: {verified}/{total}"
   

6. Print PAPER_PLAN_REPORT to terminal:

   # Paper Plan Report
   
   ## Meta
   - Title: {working title}
   - Venue: {venue}
   - Page limit: {N} pages
   - Date: {date}
   
   ## Ideas → Sections
   | Idea | Status / Novelty | Section |
   |------|------------------|---------|
   | [[primary]] | validated / 4 | Method + Exp 5.2 |
   | [[supporting-1]] | validated / 3 | Exp 5.3 |
   
   ## Page Budget
   | Section | Pages | Ideas |
   |---------|-------|-------|
   | Introduction | 1.5 | gap, contribution |
   | Related Work | 1.0 | context |
   | Method | 2.5 | primary, supporting |
   | Experiments | 2.5 | all |
   | Conclusion | 0.5 | — |
   
   ## Figures/Tables: {N} planned
   ## Citations: {verified}/{total} verified, {verify_count} [UNCONFIRMED]
   ## Review LLM Review: score {X}/10, verdict: {verdict}
   
   ## Next Steps
   - Run `/paper-draft wiki/outputs/paper-plan-{slug}-{date}.md` to draft the paper
   - Resolve {verify_count} [UNCONFIRMED] citations before /paper-compile
   

Constraints

• --venue is required: page limits and format requirements vary significantly by venue; cannot be omitted
• At least one experiment evidence: purely theoretical ideas are insufficient for an empirical paper; at least one supporting experiment is required
• Page budget must be feasible: total section pages <= venue main-body limit; otherwise adjust (compress or move to appendix)
• Review LLM review is mandatory: cannot be skipped; catching problems at the outline stage has the lowest cost
• All citations from wiki: every paper in the citation plan must exist in wiki/papers/
• idea → section mapping must be complete: every target idea must appear in at least one section
• Every section must have an idea: a section with no idea support is filler and should be removed or merged
• Graph edges via tools/research_wiki.py: do not manually edit edges.jsonl
• Citations use [[slug]]: all citations in the outline use wikilink syntax

Error Handling

• Insufficient idea status: if all ideas are proposed, error "ideas are unvalidated; run experiments first"
• No experiment evidence: error "at least one experimental result is required"; suggest running /exp-design + /exp-run first
• Insufficient wiki papers: if the citation plan has fewer than 5 wiki papers, warn "related work coverage is insufficient; consider /ingest of more papers first"
• Page budget exceeded: automatically move lower-priority sections to appendix plan; report the adjustment
• Review LLM unavailable: fall back to Claude self-review; report annotated "single-model review — cross-model verification unavailable"
• BibTeX fetch failed: mark [UNCONFIRMED]; summarize in the citation plan report
• Slug conflict: append date suffix
• Target idea not found: error; list candidates in wiki/ideas/

Dependencies

Tools（via Bash）

• python3 tools/research_wiki.py slug "<title>" — generate slug
• python3 tools/research_wiki.py add-edge wiki/ ... — add graph edge
• python3 tools/research_wiki.py rebuild-context-brief wiki/ — rebuild query_pack
• python3 tools/research_wiki.py log wiki/ "<message>" — append log
• python3 tools/fetch_s2.py search "<title>" — Semantic Scholar search (citation plan fallback)

MCP Servers

• mcp__llm-review__chat — Step 7 outline review (mandatory)

Claude Code Native

• Read — read wiki pages
• Glob — find ideas, experiments, methods, papers
• WebFetch — DBLP / CrossRef BibTeX fetch (Step 6)

Shared References

• .claude/skills/shared-references/academic-writing.md — narrative structure and section design principles
• .claude/skills/shared-references/citation-verification.md — citation fetch and verification rules

Called by

• /research Stage 5 (paper writing stage)
• Manual user invocation