Grant Writing
@K-Dense-AIK-Dense's research-grants skill: helps researchers plan, check and critique NSF, NIH, DOE, DARPA and Taiwan NSTC proposals against the actual call.
Markdown-only grant guidance with no bundled scripts; it looks up official agency pages and has an optional OpenRouter figure step in a separate skill.
Install Grant Writing in Google Antigravity
Antigravity loads skills from .agents/skills/ in your project or ~/.gemini/config/skills/ for all projects (official docs). Full guide
Install the "Grant Writing" skill for Google Antigravity. Follow the steps at https://www.antigravityskills.org/install/grant-writing.md (source: https://github.com/K-Dense-AI/scientific-agent-skills/tree/main/skills/research-grants). Before running anything, show me what you will download and whether it includes scripts, and wait for my OK. Install it into this project (.agents/skills/) unless I say global. When done, check that Antigravity lists it as an installed skill.
The agent reads the install file, shows you what it will download and waits for your OK.
Option 2: install it yourself
- 1. Download the repository
git clone --depth 1 https://github.com/K-Dense-AI/scientific-agent-skills /tmp/scientific-agent-skills && git -C /tmp/scientific-agent-skills log -1 --format=%h -- skills/research-grants
The second command prints the last commit that changed this skill. We reviewed the folder at 92ace75; a different hash means the skill changed after our review.
- 2. Copy the skill into this project (workspace skill)
mkdir -p .agents/skills && cp -r /tmp/scientific-agent-skills/skills/research-grants .agents/skills/
Run from your project root (keep references/ and assets/; SKILL.md links to them).
- 3. Or install it for all projects (global skill)
mkdir -p ~/.gemini/config/skills && cp -r /tmp/scientific-agent-skills/skills/research-grants ~/.gemini/config/skills/
Use this instead of the previous step. For the Antigravity CLI the global folder is ~/.gemini/antigravity-cli/skills/.
- 4. Check that Antigravity sees it
Antigravity 2.0 app: start a new conversation and ask "What skills do you have?"; research-grants should be listed. Antigravity CLI: run /skills (/skills reload in a session that was already open).
The README says the collection works with Google Antigravity and that npx skills add K-Dense-AI/scientific-agent-skills supports it, but that command targets the whole 177-skill collection. It also says npx skills commonly installs into ~/.agents/skills/ or .agents/skills/ and to confirm paths in your host's docs. Copying one folder, as above, adds only this skill. No API key is needed for the skill itself.
Using Claude Code, Cursor or Codex? See other assistants below.
What is Grant Writing?
How to use Grant Writing?
Key features of Grant Writing
- ✓Governing-opportunity checklist: agency, identifier and amendment, due date and time zone, eligibility, limits, attachments, review criteria, source URL and retrieval date
- ✓Requirement -> source section -> draft location -> gap matrix before any compliance verdict
- ✓5 agency guideline files: NSF, NIH, DOE, DARPA and NSTC (the NIH file is the largest agency file at 199 lines)
- ✓Guides for specific aims (464 lines), core components (404 lines) and broader impacts (398 lines)
- ✓Resubmission guidance, including responding to a summary statement
- ✓3 illustrative Markdown templates (not official agency forms)
When to use Grant Writing
- 1Extract the requirements from a NOFO or solicitation into a compliance matrix
- 2Critique an NIH R01 specific aims page against the review factors
- 3Tighten NSF broader impacts and the project summary
- 4Map reviewer concerns to revisions for a resubmission
- 5Check effort and F&A arithmetic in a budget justification
Installation & Setup Guide
Agent Rules / Prompt SetupFollow these steps (taken from the project's own docs) to configure Grant Writing in your development environment.
gh skill install K-Dense-AI/scientific-agent-skills research-grants --agent claude-code
README Option 2 (GitHub CLI v2.90.0+) shows installing one named skill and the --agent flag (claude-code, cursor, codex, gemini).
mkdir -p ~/.cursor/plugins/local && ln -s "$(pwd)" ~/.cursor/plugins/local/scientific-agent-skills
README Option 3: run from a local checkout of the repository, then reload Cursor. This loads the whole collection; gh skill install ... --agent cursor installs one skill.
npx skills add K-Dense-AI/scientific-agent-skills
README Option 1, listed for Claude Code, Claude Cowork, Codex, Gemini CLI, Google Antigravity and Cursor.
Usage Example & Verification
1) README example (from references/README.md): "Extract the requirements from this NOFO." -> What happens: the agent records the opportunity details with the source URL and builds a requirement -> source section -> draft location -> gap table. If you give it no official document and it cannot fetch one, it says so and gives a bounded review with the unresolved rules listed. 2) README example (references/README.md): "Critique my R01 aims against the review factors." -> What happens: the agent loads nih_guidelines.md, review_criteria.md and specific_aims_guide.md and returns a critique of your text. For NIH it limits itself to critique, consistency checks and light edits, because the skill says NIH excludes applications substantially developed by AI. 3) Our example: "Here are the reviewer comments from our NSF CAREER submission. Map each concern to a planned revision." -> What happens: the agent uses proposal_types_and_resubmission.md and nsf_guidelines.md to build a concern-by-concern table; it should not invent preliminary data to answer a concern.
Ask "What skills do you have?" in a new Antigravity conversation (CLI: /skills) and look for research-grants. Then paste a short funding-call excerpt and ask for a requirement matrix; the reply should list the agency, identifier, due date and source before judging anything.
Security Review
Static reviewMarkdown-only grant guidance with no bundled scripts; it looks up official agency pages and has an optional OpenRouter figure step in a separate skill.
Full review: scope, findings, risks
Static review at commit 92ace75. skills/research-grants holds SKILL.md, references/ (a map plus 11 guideline files) and assets/ (3 templates); all 16 files are Markdown and none is executable. SKILL.md frontmatter lists allowed tools Read, Write, Edit and Bash and states that the skill needs network access to verify current opportunities, or supplied dated official documents. The 77 links go to 18 hosts, mostly official agency sites. The only command shown runs python skills/scientific-schematics/scripts/generate_schematic.py from the repository root; that script belongs to another skill, and its AI path posts to https://openrouter.ai/api/v1 with OPENROUTER_API_KEY. SKILL.md itself discloses that this sends your prompt to OpenRouter. The closing 'Citing Scientific Agent Skills' section tells the agent to fetch arxiv.org or export.arxiv.org and add the K-Dense paper to your references if the skill materially contributed. No credentials are needed for normal use.
Operational Notes & Known Limitations
Known Limitations
- •Research funders only (NSF, NIH, DOE, DARPA, NSTC); there is no guidance for nonprofit, foundation or EU grants
- •Not a live opportunity database (references/README.md); facts in the files are dated (last reviewed 2026-10-01) and the agent must check the current call each time, which needs network access or documents you supply
- •For NIH it is limited to requirements, critique and light editing of investigator-written text
- •We counted: the optional figure command calls skills/scientific-schematics/scripts/generate_schematic.py "from the repository root". That script is in a different skill folder, so it does not exist if you copy only research-grants; that skill's AI script also needs the Python requests package and OPENROUTER_API_KEY
- •SKILL.md names 6 companion skills (Scientific Schematics, Scientific Writing, Literature Review, Peer Review, Research Lookup, Data Visualization); none are inside this folder
- •Difference from the technical-writing page on our site: that is a general co-writing workflow for specs and proposals; this skill is agency-specific and checks drafts against a funding call
Frequently Asked Questions
Is the grant writing skill safe?
In our static review of commit 92ace75 the folder is Markdown only, with no scripts. The agent looks up official agency pages and may fetch arxiv.org for a citation. An optional figure step in another skill sends your prompt to OpenRouter and needs OPENROUTER_API_KEY.
How do I install it in Antigravity?
Clone K-Dense-AI/scientific-agent-skills, copy skills/research-grants into .agents/skills/ or ~/.gemini/config/skills/, then ask "What skills do you have?" (CLI: /skills).
Will it write my whole NIH application?
No. The skill says NIH excludes applications substantially developed by AI, so for NIH it critiques, checks and lightly edits text you wrote.
Does it know which grants are open now?
No. It is not an opportunity database; the agent must check the current call and notices for each task.
Will it add something to my reference list?
Possibly. Its last section asks the agent to cite K-Dense's Scientific Agent Skills paper if the skill materially contributed, and to tell you it did so. Remove it if you do not want it.
Supported Platforms
Keep exploring this topic
Browse the full Antigravity skills directory for more source-linked skills with install steps and static safety reviews.
Related Skills
Legal Drafting (Contract Review)
@anthropics
Anthropic's review-contract skill: check a contract against your playbook, flag GREEN/YELLOW/RED clauses and draft ready-to-insert redlines.
Technical Writing
@anthropics
Anthropic's doc-coauthoring skill: a three-stage workflow in which the agent interviews you, drafts a spec or design doc section by section, then tests it on a fresh reader.
Skills Search (find-skills)
@vercel-labs
Vercel's find-skills skill: your AI agent searches the open skills ecosystem with npx skills find, vets results and installs the one you pick.
Popular Skills
View All →Agent Manager
@sickn33
Manage multiple local CLI agents via tmux sessions.
Systematic Debugging
@sickn33
Debug systematically, not randomly. Find and fix bugs methodically.
Browser Automation (Playwright)
@sickn33
Expert browser automation for cleaning scraping and reliable testing.