Paper Positioning Planner

Decide what the paper is selling, to whom, against which closest work, and with what evidence.

Use this skill when:

• a project has enough idea/literature/evidence to ask what the paper should be
• results are mixed and the contribution type may need to change
• a draft feels unfocused or overclaims beyond evidence
• figure review or reviewer simulation suggests the paper story is wrong
• the user needs a primary claim, secondary claims, title/abstract direction, intro thesis, or related-work boundary
• the paper may be a method paper, theory-guided method, empirical analysis, benchmark, diagnostic paper, systems paper, or negative/limitation paper
• the user is deciding between target audiences or venues before polishing the text

Do not use this skill as a paragraph-level writing adapter. Use conference-writing-adapter after the positioning decision is clear.

Pair this skill with:

• research-idea-validator when the whole project may need pursue/revise/park/kill
• literature-review-sprint when closest work or community framing is unclear
• algorithm-design-planner when the chosen position changes the method specification
• baseline-selection-audit when the position depends on whether comparisons are reviewer-proof
• figure-results-review when visual evidence changes claim scope
• paper-evidence-board when positioning decisions must update claim/evidence/provenance/risk/action/handoff links
• paper-reviewer-simulator after positioning to stress-test the selected story
• conference-writing-adapter after positioning to rewrite sections for the target venue
• research-project-memory when positioning decisions should persist across sessions

Skill Directory Layout

<installed-skill-dir>/
├── SKILL.md
└── references/
    ├── audience-venue-fit.md
    ├── contribution-claim-map.md
    ├── decision-rules.md
    ├── memory-writeback.md
    ├── narrative-architecture.md
    ├── positioning-taxonomy.md
    └── report-template.md

Progressive Loading

• Always read references/positioning-taxonomy.md, references/contribution-claim-map.md, and references/decision-rules.md.
• Read references/audience-venue-fit.md when target venue, audience, or community framing matters.
• Read references/narrative-architecture.md when producing title, abstract, intro, related-work, or main-figure direction.
• Read references/report-template.md before writing the final positioning report.
• Read references/memory-writeback.md when the project has memory/, component .agent/ folders, or the user asks for persistent memory.
• If the position depends on current venue expectations or recent accepted papers, verify with current sources, OpenReview, proceedings, or user-provided exemplars.

Core Principles

• Positioning is a decision, not a list of possible stories.
• The primary contribution must be supported by the strongest evidence, not by the user's favorite idea.
• A smaller true claim is stronger than a broad brittle claim.
• Secondary contributions should reinforce the primary story, not compete with it.
• Closest work defines the novelty boundary and the reviewer attack surface.
• The title, abstract, intro thesis, main figure, and result table should all sell the same core story.
• Claims to avoid are as important as claims to emphasize.
• A positioning decision should route concrete changes to writing, experiments, figures, or method design.

Step 1 - Recover Project State

Collect:

• one-sentence project idea or current paper thesis
• target venue or audience, if known
• current paper draft, outline, abstract, title, figures, or result tables
• literature map and closest-work risks
• baseline audit or missing-comparison risks
• figure/results review outcomes
• reviewer simulation or real review concerns
• available evidence and unsupported claims
• project memory IDs such as CLM-###, EVD-###, RSK-###, ACT-###, FIG-###, or TAB-###

Write the current story as:

This paper sells [primary contribution] to [audience] by showing [evidence] against [closest work], while avoiding the claim that [unsupported overclaim].

If the sentence cannot be written, the likely decision is revise-positioning.

Step 2 - Choose Paper Archetype

Read references/positioning-taxonomy.md.

Choose one primary archetype:

• method paper
• theory-guided method
• empirical analysis
• benchmark or dataset
• systems or tooling
• application paper
• diagnostic or mechanistic study
• negative result or limitation paper
• position or perspective paper
• hybrid paper

State why other plausible archetypes are weaker. Do not let the paper be a vague hybrid unless the evidence truly supports two linked contributions.

Step 3 - Map Contributions to Evidence

Read references/contribution-claim-map.md.

Create:

• primary contribution
• secondary contributions
• claims to keep
• claims to narrow
• claims to cut
• evidence required for each claim
• evidence currently available
• closest-work distinction
• reviewer risk if the claim stays

Every primary claim must have at least one strong evidence route. If no route exists, revise the paper archetype or route to more experiments.

Step 4 - Decide Audience and Venue Fit

Read references/audience-venue-fit.md when relevant.

Decide:

• who should care first: method researchers, theorists, benchmark users, systems builders, application researchers, or empirical analysts
• which community's standards define novelty and evidence
• whether the target venue is compatible with the strongest story
• what the paper should not try to satisfy
• what related-work boundary is needed to prevent reviewer confusion

If the evidence fits a different audience better than the user's target venue, say so directly and give the least disruptive repositioning.

Step 5 - Select the Strategic Position

Read references/decision-rules.md.

Choose exactly one:

• lock-position: story is coherent; proceed to writing adaptation
• revise-positioning: core contribution remains, but title/abstract/claims/figures must shift
• narrow-claim: evidence supports a smaller paper than the current draft claims
• change-archetype: paper type should change, such as method to empirical analysis or diagnostic study
• need-evidence: positioning depends on missing experiment, baseline, figure, theorem, or literature check
• park-paper: current evidence does not support a viable paper story yet

Do not choose lock-position if closest-work, baseline, or figure evidence risks remain fatal.

Step 6 - Build Narrative Architecture

Read references/narrative-architecture.md.

Produce:

• candidate title direction
• one-sentence thesis
• abstract skeleton
• intro paragraph roles
• main figure or main table role
• result-section ordering
• related-work boundary
• limitations to state proactively
• claims to avoid

This should be strategic and section-level. Use conference-writing-adapter later for paragraph-level venue writing.

Step 7 - Route Changes

Route every unresolved issue:

• conference-writing-adapter: position is clear and text needs venue-specific rewriting
• paper-evidence-board: claims/evidence/figures/risks must be synchronized
• figure-results-review: main figure/table does not support the selected story
• baseline-selection-audit: selected story needs stronger comparison defense
• experiment-design-planner: missing evidence must be planned
• result-diagnosis: negative or mixed results threaten the position
• literature-review-sprint: closest-work boundary remains unclear
• algorithm-design-planner: method needs to change to fit the selected claim

Step 8 - Write the Positioning Report

Read references/report-template.md.

If saving to a project and no path is given, use:

docs/paper/positioning_plan_YYYY-MM-DD_<short-name>.md

The report must include:

• current story diagnosis
• selected paper archetype
• positioning decision
• primary and secondary contributions
• claim/evidence map
• closest-work and audience boundary
• narrative architecture
• claims to avoid
• routed actions and next skills
• memory update section

Step 9 - Write Back to Project Memory

Read references/memory-writeback.md when memory exists.

Update the smallest useful set of entries:

• memory/decision-log.md: selected paper position, archetype, target audience, and revisit triggers
• memory/claim-board.md: claims kept, narrowed, revised, cut, or blocked
• memory/evidence-board.md: evidence required by the selected story
• memory/risk-board.md: positioning, closest-work, overclaim, audience, and evidence risks
• memory/action-board.md: writing, figure, experiment, baseline, or literature actions
• paper/.agent/: title/abstract/main-figure/section positioning notes

Use certainty labels:

• verified for evidence checked against results, draft text, or sources
• user-stated for user goals and constraints
• inferred for strategic judgments and reviewer-risk predictions
• unverified for positions depending on unchecked current literature or missing results

Final Sanity Check

Before finalizing:

• one primary story is selected
• paper archetype is explicit
• audience and closest-work boundary are clear
• every primary claim has evidence or a routed action
• unsupported claims are named and removed/narrowed
• title/abstract/main-figure direction match the same story
• next skill is unambiguous
• project memory is updated when present