Workflow Planning & Writing Skills

Orchestrates the end-to-end analysis workflow from planning through implementation and review to finalized results. Chains analysis-planning, code-writing, and code-review skills in sequence with state tracking and user checkpoints. Use this skill when the user wants to run the full analysis workflow, resume from a checkpoint, check pipeline status, or coordinate between planning, coding, and review phases. Trigger phrases include: "run the full analysis workflow", "where are we in the analysis", "review this code", "run another review round", "finalize the analysis", or "resume the analysis from where we left off". The orchestrator manages state, tracks completion, and coordinates handoffs between skills. Each component skill remains independently usable.

Researches statistical approaches and published examples to produce a structured analysis plan for scientific data. Use this skill when the user mentions: planning an analysis, choosing a statistical model, designing an analytical approach, deciding between model types, preparing for a new analysis, or wanting to understand which methods to use. Also trigger when users describe a research question and dataset and want help selecting appropriate models. The skill runs parallel sub-agents to research fundamentals, find published examples, and assess detection or measurement bias (if applicable), then synthesizes into a justified plan with diagnostics checklist. Works standalone or as the first step in the analysis-pipeline.

Converts a markdown manuscript (.md) into a journal-ready Word document (.docx) with proper formatting, citation styles, and optional tracked-change conversion. Use this skill whenever the user wants to: convert a markdown manuscript to Word format, render a manuscript for journal submission, build a .docx from a markdown draft, apply journal-specific formatting to a manuscript, or re-render a manuscript after edits. Also trigger when the user says "build the final docx", "format for submission", or "convert to Word". Works standalone or as the final step in the manuscript-pipeline workflow.

Orchestrates the end-to-end manuscript workflow from research through review to submission-ready .docx. Chains paper-research, expert-review, and manuscript-builder skills in sequence. Use this skill when the user wants to run the full pipeline, resume from a checkpoint, or coordinate between research, review, and formatting phases. Trigger phrases include: "run the full manuscript workflow", "where are we in the pipeline", "start manuscript review", "run another review round", "build the final docx", or "finalize the manuscript". The orchestrator manages state, tracks completion, and coordinates handoffs between skills. Each component skill remains independently usable.

Lightweight entry point for ad-hoc manuscript writing: drafting, refining, or editing journal manuscript text without running a full pipeline. Use whenever the user wants to draft a paragraph, refine an introduction, improve a sentence, write methods, expand a discussion, edit for voice, or apply the lab's writing conventions to existing prose. Trigger phrases: "help me draft this paragraph", "refine my introduction", "improve this sentence", "write the methods section", "edit for voice", "polish this paragraph", "tighten this writing", "fix the hedging in this sentence". Lightweight counterpart to the paper-research workflow (multi-phase pipeline). Route to paper-research when literature search is needed, expert-review for peer-review simulation, or manuscript-builder for final .docx rendering.

End-to-end AI-assisted workflow for taking analysis findings and producing a publication-ready manuscript reference document. Use this skill whenever the user mentions: writing a paper, preparing a manuscript, needing background literature, filling in an introduction or discussion section, turning results into a publishable paper, or needing to contextualize research findings within existing literature. Also trigger when users ask for help synthesizing literature for any section of a scientific paper, or want to prepare context documents for manuscript drafting. The skill handles everything from theme extraction through literature synthesis, contradiction-checking, and structured reference document generation: the user should NOT need to manually find or upload PDFs.

Lightweight entry point for ad-hoc reviewer reply writing: drafting, refining, or editing reply text and manuscript inserts without running the full pipeline. Use whenever the user wants to draft a reply to a single reviewer comment, refine an existing reply, improve tone or push-back wording, check voice on a reply, write a manuscript insert prompted by a comment, or apply the lab's reply conventions. Trigger phrases: "draft a reply to this comment", "help me respond to R1-C3", "refine this reply", "check the voice on this reply", "edit my reviewer response", "tighten this reply". Lightweight counterpart to reviewer-reply-planning/drafting (multi-comment pipeline with parallel sub-agents and tracked-changes rendering). Route to reviewer-reply-planning for full triage, or reviewer-reply-drafting for end-to-end tracked-changes render.

Iteratively refines a research analysis through multi-specialist critique and quality gates until it reaches publication-ready quality. Use this skill when the user says: "iterate the analysis", "refine until publication-ready", "run through quality gates", "bring this to publication quality", or asks for a defensible, multi-round refinement of existing outputs. The skill orchestrates the Lab Director + specialist agents through a six-phase loop (readiness assessment, planning, implementation, parallel critique, arbitration, and self-update) until the project's concrete goal spec is met, escalating to the user only on structural decisions. Wraps analysis-pipeline as a sub-workflow for code-heavy rounds; adds cross-domain critique, render-and-read visualization review, and self-improvement feedback into the framework itself.

Drafts the final reviewer reply document and applies tracked-change edits to the manuscript, given a frozen plan from reviewer-reply-planning. Use this skill whenever the user mentions: drafting the reply to reviewers, building the response document, applying tracked changes from the reply plan, finalizing the reviewer response, or rendering the revised manuscript with tracked changes. Also trigger when users say "draft the reply now", "build the response doc", or "apply the planned edits to the manuscript". Requires a [title]replyplan.md as input: if no plan exists, route to reviewer-reply-planning first. Produces two .docx outputs (reply doc + revised manuscript with tracked changes) via manuscript-builder. Supports targeted re-entry on individual comments without redrafting the whole reply.

Orchestrates the end-to-end reviewer-reply workflow from raw reviewer comments to a submission-ready response document and tracked-changes manuscript. Chains reviewer-reply-planning and reviewer-reply-drafting in sequence, manages state, and coordinates handoffs. Use this skill whenever the user wants to run the full reply workflow, resume from a checkpoint, redraft individual comments, or check status. Trigger phrases include "address the reviewer comments", "respond to the reviewers", "where are we on the reply", "redraft my reply to R2-C3", "build the final reply docx", or "resume the reply workflow". The orchestrator manages a JSON state file alongside the manuscript and supports targeted re-entry on individual comments without redrafting the whole reply.

Parses reviewer comments and a manuscript, classifies every comment by action type, and produces a triage plan with proposed edits and replies. Use this skill whenever the user mentions: addressing reviewer comments, planning a manuscript revision, classifying review feedback, triaging reviewer comments, building a reply plan, or starting the reviewer-reply workflow. Also trigger when users upload a manuscript .docx alongside a reviewer comments .docx and ask what to do with them. This skill produces a structured plan: it does NOT yet draft the final reply or apply edits to the manuscript. Use reviewer-reply-drafting next, or run the full reviewer-reply-pipeline. The planning step front-loads the user's input via AskUserQuestion on comments needing judgement (RESTRUCTURE, NEW-ANALYSIS, JUDGEMENT) so the drafting skill can run without further interruption.

Research a topic and write about it: syntheses, literature reviews, perspectives, policy briefs, technical reports, blog posts, or background documents. Triggers: "write a synthesis on", "research and summarize", "write a perspective piece", "draft a policy brief", "write a blog post about", "put together a review of", "what does the literature say about". Not for journal manuscripts from analysis findings (use paper-research) or manuscript review (use expert-review).