grants

Imported from borghei/claude-skills (research/grants/SKILL.md). Install upstream with npx skills add borghei/claude-skills --skill grants. Copyright stays with the author (MIT + Commons Clause).

Grant Writing & Proposal Architecture

A skill for crafting competitive grant proposals across funder types: government (NIH, NSF, DOE, ARPA), foundation, corporate, philanthropic, and SBIR / STTR. Focuses on the architecture of a winning proposal: fit, structure, narrative, budget — not boilerplate templating.

When to use this skill

• Evaluating funder fit before investing weeks in a proposal
• Designing the proposal structure for a specific funder
• Writing or auditing the narrative for competitiveness
• Designing a realistic, defensible budget
• Pre-submission proposal review for common failure modes
• Building a grants strategy (which to apply to over the year)

Inputs the advisor expects

• The funder name + specific program / RFP
• The research / project idea (problem, approach, outcomes)
• Team composition (PI, co-investigators, key personnel)
• Institutional / org context
• Past funding history
• Budget envelope (or constraint)
• Submission deadline

Clarify First

Before generating the proposal, settle the inputs below. When one is missing or vague and would change the result, ask about it; when it wouldn't, use a sensible default and say which one you used.

• Funder + specific program / RFP: sets the mental model (NIH 5-criteria vs NSF merit+impact vs foundation mission fit); drives structure and narrative.
• Project idea (problem, approach, outcomes): drives the significance/innovation narrative and the Heilmeier answers.
• Budget envelope: drives budget design and whether scope matches the funder's typical award size.
• Team composition (PI, key personnel): drives the investigator/environment fit dimension.

Ask about the two or three inputs that matter most for this request, not the whole list. If the user wants a result straight away, produce it and state your assumptions at the top.

Workflows

Workflow 1 — Score funder fit before committing

1. Capture funder, program, project idea, team strengths.
2. Run funder_fit_scorer.py to grade fit on 7 dimensions.
3. If fit < 65, look for better-aligned funder; don't waste 4 weeks.

python3 grants/scripts/funder_fit_scorer.py \
  --input funder_fit.json --format markdown

Workflow 2 — Validate proposal structure against funder expectations

1. Capture the proposal section list + page allocation.
2. Run proposal_structure_validator.py against funder type expectations.
3. Adjust before drafting deep.

python3 grants/scripts/proposal_structure_validator.py \
  --input proposal_structure.json --funder-type nih --format markdown

Workflow 3 — Audit budget for realism

1. Capture budget line items with justifications.
2. Run budget_realism_checker.py against funder norms + project scope.
3. Adjust before submission.

python3 grants/scripts/budget_realism_checker.py \
  --input budget.json --format markdown

Decision frameworks

Funder fit dimensions

1. Topic alignment — does the funder fund this area?
2. Mechanism alignment — does the funder fund this kind of work (R&D, services, scale-up)?
3. Stage alignment — early-stage / mid / scale?
4. Geographic alignment — does the funder fund your region?
5. Team profile alignment — does the funder fund your kind of team?
6. Budget envelope alignment — does the funder's typical award size match?
7. Competitive density — is it 5% acceptance or 35%?

A score below 65 across these is usually a "skip this funder" signal.

Funder type — distinct mental models

Write to the funder's mental model, not a generic "good grant."

The Heilmeier catechism (good for any proposal)

1. What are you trying to do?
2. How is it done today; what are the limits?
3. What's new in your approach; why succeed?
4. Who cares; if you succeed, what difference does it make?
5. What are the risks; how will you mitigate?
6. How much will it cost; how long?
7. What are the mid-term + final outcomes you'll deliver?

A proposal that can't answer all seven crisply isn't ready.

Common engagements

"Help me decide between two RFPs"

1. Score both for fit; the higher one is usually right.
2. If close: which has earlier deadline / smaller proposal effort?
3. Don't submit to both same year unless funders are independent.

"Audit my draft proposal"

1. Check funder-fit assumptions (did the program actually fund what you're proposing?)
2. Check structure against funder template
3. Check narrative: is the problem compelling? approach novel?
4. Check budget: realistic + justified
5. Check team credentials: matches scope?
6. Read for: jargon, vague claims, unjustified assumptions

"We've never applied for an NIH R01. What's the prep?"

1. Smaller grant first (R21, K, F32) if eligible — build track record
2. Talk to a program officer before drafting (essential)
3. Pre-submission inquiry where allowed
4. Get a mock review from someone who's reviewed for NIH

Anti-patterns to avoid

• Applying without funder fit. Wastes 4-8 weeks.
• Generic proposal sent to multiple funders. Each wants a specific mental model.
• Budget that doesn't match scope. Reviewer red flag.
• Vague significance statement. "This is important" without specifics.
• No risk discussion. Reviewers know there's risk; not acknowledging it = naive.
• Team without right credentials. Match key personnel to scope.
• Submitting at last minute. Errors; missed letters of support.

Scope & Limitations

In Scope:

• Funder fit scoring across seven dimensions (topic, mechanism, stage, geography, team profile, budget envelope, competitive density) with funder_fit_scorer.py
• Proposal structure checks against funder-type expectations (NIH, NSF, SBIR, foundation, DARPA, generic) with proposal_structure_validator.py: missing sections and page-limit problems
• Budget realism checks with budget_realism_checker.py: personnel effort, PI effort sum, equipment and supplies balance, travel, indirect rate and justification coverage
• Narrative guidance using funder mental models and the Heilmeier catechism

Out of Scope:

• The literature review behind the proposal (see research/litreview/)
• Detailed study costing from unit costs (see research-ops/research-finance/)
• Legal review of award terms (see c-level-advisor/general-counsel-advisor/)
• Searching funder databases for open calls and submitting through funder portals

Important Caveats:

• Fit ratings such as mechanism or budget alignment are judgments the user enters. The score summarizes them and does not check them against the funder.
• Section lists and page limits built into the validator are general patterns per funder type. The current solicitation and funder guide always take precedence, and they change between cycles.
• Indirect cost rates and allowable costs depend on the institution and the funder. A clean run does not mean the proposal will be funded.

References

• references/funder-fit-and-research-strategy.md — fit dimensions, funder types, multi-funder strategy
• references/proposal-structure-and-narrative.md — per-funder structures, narrative discipline
• references/budget-design-and-justification.md — budget categories, indirect costs, common errors

Related skills

• research/litreview — literature review for proposals
• c-level-advisor/general-counsel-advisor — legal review of terms
• c-level-advisor/cfo-advisor — financial review