papermill

Draft a paper using multi-agent writing system

Design experiments or computational studies

Initialize or refresh a paper repo for papermill

Design or refine the paper structure

Final pre-submission preparation checklist

Conduct a systematic literature survey

Develop or verify mathematical proofs

Get structured editorial feedback on a draft

Design Monte Carlo simulations

Show paper project status dashboard

Extract or refine the paper's central claim

Find and evaluate publication venues

Bibliographic specialist verifying citation accuracy, relevance, completeness, and related work fairness. Uses web search to verify citations exist and match claims. Launched by the reviewer orchestrator. <example> Context: Orchestrator needs citation verification. user: "Verify all citations in this paper" assistant: "I'll launch the citation-verifier to check every reference for accuracy and relevance." </example> <example> Context: Paper's related work section needs scrutiny. user: "Is the related work section fair and complete?" assistant: "I'll launch the citation-verifier to evaluate related work coverage and citation accuracy." </example>

Specialist writer for mathematical content — definitions, theorems, proofs, derivations, and formal analysis. Produces rigorous LaTeX theorem environments with clear proof structure. Launched by the writer orchestrator during multi-agent drafting. <example> Context: Orchestrator needs theory sections drafted. user: "Write the main results section with proofs" assistant: "I'll launch the formal-writer to draft theorems, proofs, and derivations." </example> <example> Context: Paper needs a preliminaries section with definitions. user: "Draft the preliminaries with notation and key definitions" assistant: "I'll launch the formal-writer to establish the mathematical framework." </example>

Manuscript production specialist ensuring the paper builds cleanly and meets venue formatting requirements. Checks LaTeX/Markdown/RMarkdown compilation, label resolution, BibTeX integrity, and venue compliance. Launched by the reviewer orchestrator. <example> Context: Orchestrator needs build and format verification. user: "Check if this paper builds correctly and meets formatting requirements" assistant: "I'll launch the format-validator to verify compilation and venue compliance." </example> <example> Context: Paper needs pre-submission formatting check. user: "Does this paper meet the formatting requirements for the target venue?" assistant: "I'll launch the format-validator to check venue compliance and build integrity." </example>

Field surveyor that maps the landscape of competing and related approaches for a paper under review. Searches broadly at multiple levels of specificity to find competing approaches, benchmarks, and state-of-the-art comparisons. Launched by the reviewer orchestrator. <example> Context: Orchestrator needs broad literature context for review. user: "Survey the field around this paper's topic" assistant: "I'll launch the broad literature scout to map competing and related approaches." </example> <example> Context: Paper needs contextualization within its research area. user: "What is the landscape of related work for this paper?" assistant: "I'll launch the broad literature scout to find competing approaches and benchmarks." </example>

Direct comparison specialist that finds the closest related work to a paper under review. Focuses on papers addressing the exact same problem, using the same techniques, or making overlapping claims. Launched by the reviewer orchestrator. <example> Context: Orchestrator needs targeted comparisons for novelty assessment. user: "Find the closest competing work to this paper" assistant: "I'll launch the targeted literature scout to find direct comparisons and overlapping claims." </example> <example> Context: Paper's novelty needs verification against specific prior work. user: "Has anyone done exactly this before?" assistant: "I'll launch the targeted literature scout to find the most directly competing work." </example>

Specialist writer for related work and background sections. Turns literature scout findings into a compelling narrative that positions the paper's contribution against the field. Launched by the writer orchestrator during multi-agent drafting. <example> Context: Orchestrator needs a related work section drafted. user: "Write the related work section positioning our contribution" assistant: "I'll launch the literature-writer to draft a narrative related work section." </example> <example> Context: Paper needs a background section synthesizing prior approaches. user: "Draft the background section covering existing methods" assistant: "I'll launch the literature-writer to synthesize prior work into a coherent background." </example>

Rigorous mathematical and logical reviewer for academic papers. Checks proof correctness, assumption sufficiency, logical chain integrity, and claim support. Launched by the reviewer orchestrator during multi-agent review. <example> Context: Orchestrator needs proof and logic verification. user: "Check all proofs and logical arguments in this paper" assistant: "I'll launch the logic-checker agent to verify every proof step and logical chain." </example> <example> Context: Paper has multiple theorems that need verification. user: "Verify the mathematical correctness of this draft" assistant: "I'll launch the logic-checker agent to systematically check each theorem and derivation." </example>

Specialist writer for methodology, algorithms, and experimental design sections. Writes clear technical procedures with reproducibility in mind. Launched by the writer orchestrator during multi-agent drafting. <example> Context: Orchestrator needs a methodology section drafted. user: "Write the experimental methodology section" assistant: "I'll launch the method-writer to draft the methodology with reproducibility details." </example> <example> Context: Paper needs an algorithm description. user: "Draft the algorithm design section with pseudocode" assistant: "I'll launch the method-writer to describe the algorithm and its design rationale." </example>

Evaluates experimental design, statistical rigor, reproducibility, and baseline appropriateness in academic papers. Checks whether results can be trusted and reproduced. Launched by the reviewer orchestrator. <example> Context: Orchestrator needs methodology evaluation. user: "Audit the experimental methodology in this paper" assistant: "I'll launch the methodology-auditor to evaluate design, statistics, and reproducibility." </example> <example> Context: Paper has computational experiments that need scrutiny. user: "Are the experiments in this paper sound and reproducible?" assistant: "I'll launch the methodology-auditor to check experimental rigor." </example>

Assesses whether a paper's contributions are genuine advances over prior work. Evaluates contribution clarity, differentiation, significance, and originality using literature context. Launched by the reviewer orchestrator. <example> Context: Orchestrator needs novelty evaluation with literature grounding. user: "Assess whether this paper's contributions are novel" assistant: "I'll launch the novelty-assessor to evaluate contributions against the literature." </example> <example> Context: Paper claims a new method but similar work may exist. user: "Is this contribution genuinely new or incremental?" assistant: "I'll launch the novelty-assessor to compare claims against known prior work." </example>

Experienced academic editor evaluating writing quality, narrative structure, notation consistency, and communication effectiveness in research papers. Launched by the reviewer orchestrator. <example> Context: Orchestrator needs writing quality evaluation. user: "Evaluate the writing quality of this paper" assistant: "I'll launch the prose-auditor to assess clarity, structure, and notation consistency." </example> <example> Context: Paper may have unclear exposition. user: "Is this paper well-written and clearly organized?" assistant: "I'll launch the prose-auditor to check narrative arc and prose quality." </example>

Specialist writer for results, analysis, and discussion sections. Presents findings clearly, interprets them honestly, and discusses implications and limitations. Launched by the writer orchestrator during multi-agent drafting. <example> Context: Orchestrator needs results written up from experimental data. user: "Write the results section presenting our findings" assistant: "I'll launch the results-writer to present findings with analysis and discussion." </example> <example> Context: Paper needs a discussion section interpreting results. user: "Draft the discussion covering implications and limitations" assistant: "I'll launch the results-writer to interpret results and discuss limitations." </example>

Multi-agent review orchestrator. Acts as area chair: reads the paper, spawns literature scouts (parallel), then specialist reviewers (parallel), cross-verifies critical findings, and synthesizes a unified report to .papermill/reviews/. Launch from /papermill:review for thorough autonomous analysis. <example> Context: User wants a deep autonomous review of their paper draft. user: "Do a thorough review of my paper" assistant: "I'll launch the reviewer agent for a comprehensive multi-agent editorial review." </example> <example> Context: User wants every theorem and equation checked systematically. user: "Check all the proofs and equations in my paper" assistant: "I'll launch the reviewer agent to systematically verify every theorem and equation." </example>

Autonomous literature survey agent. Systematically searches academic databases, follows citation networks, and compiles a structured reference list with classifications and gap analysis. Launch from /papermill:prior-art for deep coverage. <example> Context: User wants a broad autonomous literature search beyond the interactive survey. user: "Do a deep literature search on this topic" assistant: "I'll launch the surveyor agent for a systematic autonomous literature survey." </example> <example> Context: User wants to explore citation networks and find more references. user: "Find more related papers by following citation networks" assistant: "I'll launch the surveyor agent to follow citation networks and compile an extended reference list." </example>

Multi-agent writing orchestrator. Acts as lead author: reads context, plans the paper, spawns literature scouts and section writers in parallel, integrates drafts into a unified manuscript, and writes bookend sections (abstract, introduction, conclusion). Launch from /papermill:draft for autonomous paper drafting. <example> Context: User wants to bootstrap a full paper draft from an outline. user: "Draft my paper based on the outline" assistant: "I'll launch the writer agent for multi-agent paper drafting." </example> <example> Context: User has thesis and outline, needs the manuscript written. user: "Write the paper" assistant: "I'll launch the writer agent to coordinate specialist writers and produce a full draft." </example>

This skill should be used when the user asks to "draft my paper", "write the paper", "bootstrap a draft", "generate a manuscript", "write the first draft", "create a paper draft", or needs to produce a full or partial manuscript draft. Launches a multi-agent writing system with specialist writers for literature, formal content, methodology, and results, orchestrated by a lead author agent. Produces a manuscript draft and section-level artifacts in .papermill/drafts/. Updates .papermill/state.md.

This skill should be used when the user asks to "design an experiment", "plan my experiments", "set up a benchmark", "how should I test my thesis", "design a computational study", or needs to plan experiments for a research paper. Covers hypothesis formulation, variable identification, methodology selection, and success criteria definition. Produces a structured experiment plan with reproducibility in mind.

This skill should be used when the user asks to "initialize a paper", "start a new paper project", "set up papermill", "onboard this repo", "refresh papermill", "update papermill state", or needs to create or update the .papermill/state.md state file. On new repos: discovers structure, infers format, gathers author info, creates state file. On existing repos: offers refresh mode that adds missing schema fields and new context without overwriting existing data. Migrates old .papermill.md files to .papermill/state.md automatically.

This skill should be used when the user asks to "outline my paper", "structure the paper", "design the paper sections", "create a paper outline", "organize my argument", or needs to design or refine paper structure. Generates a section-by-section outline with purpose, key arguments, estimated length, and narrative arc. Adapts to the paper type and venue conventions. Updates .papermill/state.md.

This skill should be used when the user asks to "polish my paper for submission", "pre-flight check", "is my paper ready to submit", "final check before submission", "prepare for submission", or needs final preparation before submitting a paper. Performs a systematic pre-flight check covering formatting, citations, figures, metadata, and build verification. Produces a checklist report with suggested fixes.

This skill should be used when the user asks to "survey related work", "find prior art", "literature review", "what papers are related to mine", "search for references", or needs to conduct a systematic literature survey. Reads thesis from .papermill/state.md, searches academic sources, classifies references, identifies gaps, generates BibTeX, and updates the state file. Can launch the surveyor agent for deep autonomous search.

This skill should be used when the user asks to "help me prove this theorem", "check my proof", "verify this proof is correct", "develop a proof for", "is this proof rigorous", or needs to develop, verify, or refine mathematical proofs. Guides proof strategy selection, checks logical correctness, identifies gaps, and helps with clear mathematical writing. Works with existing proof drafts or constructs from scratch.

This skill should be used when the user asks to "review my paper", "give me feedback on my draft", "editorial review", "is my paper ready to submit", "check my paper for issues", or needs structured editorial feedback on a paper draft. Launches a multi-agent review system with 8 specialists (2 literature scouts + 6 domain reviewers) orchestrated by an area chair agent. Produces a unified report in .papermill/reviews/. Updates .papermill/state.md.

This skill should be used when the user asks to "design a Monte Carlo simulation", "validate my theoretical result", "set up a simulation study", "how many replications do I need", "plan a coverage study", or needs to design simulations for verifying theoretical results. Covers sample size determination, convergence diagnostics, and result presentation. Specialized for statistical methodology papers.

This skill should be used when the user asks to "show paper status", "where am I on this paper", "what's the state of my paper", "papermill dashboard", or needs orientation on a paper project. Reads .papermill/state.md and presents a read-only dashboard showing stage, thesis, experiments, review history, venue, and suggested next actions.

This skill should be used when the user asks to "define my thesis", "sharpen my research question", "what is my paper's main claim", "extract the thesis from my draft", "crystallize my contribution", or needs to refine the central claim and novelty of a research paper. Works in two modes: extracting the thesis from an existing draft, or Socratic dialogue to sharpen a new idea. Updates .papermill/state.md.

This skill should be used when the user asks to "where should I submit this paper", "find a venue for my paper", "which journal fits my paper", "evaluate publication venues", "conference or journal for this work", or needs to identify and evaluate publication venues. Considers scope, impact factor, review timeline, acceptance rate, and fit. Produces a ranked shortlist with submission strategy guidance.