nature-proposal-writer

Drafts and revises research proposals through evidence-first foundation files and four-gate QA.

2|Updated Jun 18, 2026
One-click install
npx skills add https://github.com/AlexenderSokolov/deepscientist-lite-codex-plugin --skill nature-proposal-writer-alexendersokolov
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: nature-proposal-writer
Source: https://github.com/AlexenderSokolov/deepscientist-lite-codex-plugin/tree/main/plugins/deepscientist-lite-academic/skills/nature-proposal-writer
Command: npx skills add https://github.com/AlexenderSokolov/deepscientist-lite-codex-plugin --skill nature-proposal-writer-alexendersokolov

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires python-docx, and includes scripts (resource) and references (resource) components.

What problem does it solve? Research proposals and academic drafts often suffer from unsupported claims, weak argument structure, and template-like language. This Skill enforces an evidence-before-prose workflow so every claim is traceable, every section has a contract, and quality is scored against explicit rubrics before delivery. ## Core Features & Use Cases - Three writing modes: compose (from a topic or idea), revise (existing paragraphs or full proposals), and hybrid (preserve a draft while expanding missing structure). - Foundation file system: builds research canon, evidence tables, argument maps, section contracts, and style guides before any prose is drafted. - Four-gate QA pipeline: dynamic expert review, anti-slop language scan, automated validation of citations and reproducibility, and 8-dimension rubric scoring with thresholds. - Use Case: A doctoral student provides a rough research direction on materials corrosion; the Skill builds the canon and evidence table, drafts section by section under contracts, runs QA, and exports a formatted .docx proposal. ## Quick Start Ask the assistant to use the nature-proposal-writer skill to turn your research topic into a structured proposal with an evidence table and QA report.

Frequently Asked Questions about nature-proposal-writer

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I write a research proposal from a rough topic idea?▼

Use compose mode: the workflow runs intake questions, locks scope, then builds a research canon, evidence table, argument map, and section contracts before drafting. Drafting only begins after the foundation review scores above 7.5.

How to revise an existing proposal draft for weak claims?▼

Use revise mode, which extracts claims from your text, compares them against the canon and evidence table, diagnoses structure, and produces a prioritized revision brief. Unsupported claims are downgraded or flagged, never silently upgraded.

What is the difference between compose, revise, and hybrid modes?▼

Compose builds a proposal from a topic or idea, revise diagnoses and fixes existing text, and hybrid preserves an existing draft while composing missing sections. Hybrid runs revise first, then composes the gaps, then merges and runs full QA.

Does the proposal export support DOCX format?▼

Yes, the included build_proposal_docx.py script converts Markdown to DOCX using python-docx with Times New Roman and SimSun fonts, 1.5 line spacing, and 1-inch margins. Pandoc export is explicitly disallowed due to poor formatting.

When should I stop iterating on proposal revisions?▼

Stop after 3 revision rounds, when two consecutive score improvements are under 0.5, when key evidence is missing, or when expert conflicts cannot be resolved. The workflow reports remaining risks and one next action instead of polishing over gaps.

What are the limitations of automated proposal writing?▼

The skill never invents experimental results, references, or feasibility evidence; missing facts are marked AUTHOR_INPUT_NEEDED. For sentence-level English polishing only, a dedicated polishing skill is recommended instead.