research-grants

Trigger phrases

Phrases that activate this skill when typed to Claude Code:

• write a grant proposal
• draft an NIH application
• NSF broader impacts
• write specific aims
• DARPA proposal

What it does

research-grants is a Claude Code skill from K-Dense AI’s scientific-agent-skills repo. It turns Claude into a grant-writing collaborator that knows the specific requirements, review criteria, and formatting rules for major funding agencies — NSF, NIH, DOE, DARPA, and Taiwan NSTC.

A session produces targeted proposal sections: specific aims pages, significance and innovation narratives, research strategy sections, broader impacts, and budget justifications — each calibrated to the agency’s reviewer scoring criteria. The skill handles the agency-specific page limits, formatting requirements, and compliance details that make or break submissions.

When to use it

Reach for it when:

• You’re starting a new grant application and need to structure the narrative around the funder’s review criteria
• You have a rough research plan and need to translate it into the significance/innovation/approach structure NIH reviewers expect
• You’re writing NSF broader impacts and want language that addresses the review criteria without sounding boilerplate

When not to reach for it:

• Scientific paper writing — use scientific-writing for manuscripts
• Generating the preliminary data itself — the skill writes about data, it doesn’t produce it

Install

Copy the SKILL.md from K-Dense AI’s research-grants folder into .claude/skills/research-grants/ in your project.

Trigger phrases: “write a grant proposal”, “draft an NIH application”, “NSF broader impacts”, “write specific aims”.

What a session looks like

A typical session has three phases:

1. Agency and section scoping. Identify the target agency, funding mechanism, and which sections need drafting. Claude retrieves the current review criteria and page limits for that mechanism.
2. Section drafting. Each section is drafted with the reviewer’s scoring lens in mind — significance frames the gap, innovation articulates the advance, approach maps methods to aims with contingency plans.
3. Compliance check. Claude reviews the draft against formatting requirements (margins, font size, page count) and flags anything that would trigger an administrative return.

Receipts

Where it works well:

• Specific aims pages — the one-page structure where framing and logic density matter most; Claude produces well-organized first drafts quickly
• Agency-specific boilerplate like data management plans and resource sharing statements, where the required elements are well-defined

Where it backfires:

• Highly novel research areas where Claude’s knowledge of the current funding landscape may be out of date
• Budget preparation for complex subcontracts — the numbers need to come from your institution’s grants office

Pattern that works: write the specific aims first with this skill, then use the approved aims as the anchor for all other sections in the application.

Source and attribution

Originally authored by K-Dense Inc.. The canonical SKILL.md lives in the research-grants folder of their public scientific-agent-skills repository.

License: MIT. Install, adapt, and redistribute with attribution preserved.

This page documents the skill from a practitioner’s perspective. For the formal spec and any updates, defer to the source repo.