Source: https://github.com/aipoch/medical-research-skills

When to Use

Use this skill when you need to produce or revise a grant application that must meet strict agency rules and reviewer expectations, for example:

1. Preparing a new submission to NSF, NIH, DOE, DARPA, or Taiwan’s NSTC in response to a specific solicitation/FOA/BAA.
2. Drafting core narrative sections (NSF Project Description, NIH Research Strategy, DARPA Technical Volume, DOE Project Narrative, NSTC CM03).
3. Building agency-specific “value” sections, such as NSF Broader Impacts, NIH Significance/Innovation, or DARPA transition and milestone narratives.
4. Creating a compliant budget + justification aligned to scope, timeline, and agency constraints (e.g., NIH modular budgets, DARPA phase/task budgets).
5. Resubmitting after reviews, including structured responses to critiques (especially NIH A1) and targeted strengthening of weak criteria.

Key Features

• Agency-aware structure and compliance

  • NSF: Intellectual Merit + Broader Impacts, typical 15-page Project Description norms
  • NIH: Specific Aims + Significance/Innovation/Approach framing, rigor/reproducibility expectations
  • DOE: office-dependent emphasis (Office of Science, ARPA-E, EERE), partnerships/cost-share where applicable
  • DARPA: high-risk/high-reward framing, measurable milestones, transition pathways, phased execution
  • NSTC (Taiwan): CM03-centered technical narrative, bilingual abstract expectations, feasibility emphasis

• Review-criteria-driven writing

  • Maps every major claim to what reviewers score (or discuss) and what program staff prioritize.

• Budget-to-scope alignment

  • Ensures personnel effort, equipment, travel, subawards, and indirects match the workplan and schedule.

• Milestones, timeline, and management planning

  • Produces Gantt-style schedules, go/no-go criteria, deliverables, and risk mitigation (especially important for DARPA/DOE).

• Mandatory visual communication workflow

  • Every proposal should include at least 1–2 diagrams (e.g., workflow, conceptual framework, timeline). Use the scientific-schematics skill to generate publication-quality figures.

• Reference-driven drafting

  • Leverages the repository’s detailed guides as needed:
    • references/nsf_guidelines.md
    • references/nih_guidelines.md
    • references/doe_guidelines.md
    • references/darpa_guidelines.md
    • references/nstc_guidelines.md
    • references/specific_aims_guide.md
    • references/broader_impacts.md
    • references/budget_preparation.md
    • references/review_criteria.md
    • references/timeline_planning.md
    • references/team_building.md
    • references/resubmission_strategies.md

Dependencies

• Python: 3.10+ (recommended)
• Optional local scripts (repository-provided):
  • scripts/compliance_checker.py (format checks)
  • scripts/budget_calculator.py (budget math support)
  • scripts/deadline_tracker.py (planning support)
  • scripts/generate_schematic.py (diagram generation wrapper; used with scientific-schematics)

Note: Exact third-party Python package requirements are not specified in the source document. If you maintain this skill repository, add a requirements.txt (with pinned versions) and list them here.

Example Usage

The example below is a complete, runnable workflow that (1) generates required visuals, (2) drafts core sections, and (3) performs basic compliance checks using the included scripts.

1) Generate required diagrams (minimum 1–2)

# Conceptual framework / workflow diagram
python scripts/generate_schematic.py \
  "Conceptual workflow for a 3-aim biomedical project: Aim 1 data collection -> Aim 2 model development -> Aim 3 validation; include feedback loop and key deliverables" \
  -o figures/workflow.png

# Timeline / milestones diagram (recommended)
python scripts/generate_schematic.py \
  "Gantt chart for a 3-year project with quarterly milestones; include go/no-go at end of Year 1 and deliverables per aim" \
  -o figures/timeline.png

2) Draft an NIH-style proposal skeleton (Specific Aims + Strategy)

Create proposal.md:

# Project Title
Mechanistic and Translational Study of X to Enable Y

## NIH Specific Aims (1 page target)
**Knowledge gap:** ...
**Long-term goal:** ...
**Objective:** ...
**Central hypothesis:** ...

**Aim 1 (verb-led):** ...
- Rationale:
- Approach (high level):
- Expected outcomes:

**Aim 2:** ...
**Aim 3:** ...

**Impact:** If successful, this work will ...

## Research Strategy (12 pages target for R01)

### Significance
- Problem and barrier to progress:
- Why now / why this team:
- Expected impact on health/biology:

### Innovation
- Conceptual innovation:
- Methodological innovation:
- Why current approaches are insufficient:

### Approach
#### Overview and rationale
#### Aim 1 Methods
- Design:
- Data:
- Analysis:
- Pitfalls and alternatives:
#### Aim 2 Methods
...
#### Aim 3 Methods
...

### Rigor and Reproducibility (as applicable)
- Controls, replicates, blinding/randomization:
- Power/statistics:
- Data management and sharing:

3) Run a basic formatting/compliance check (if available)

python scripts/compliance_checker.py proposal.md

4) Produce a budget justification draft (outline)

Create budget_justification.md:

# Budget Justification (Draft)

## Personnel
- PI (X% effort): ...
- Postdoc (100%): ...
- Graduate student (50%): ...

## Equipment
- Item: purpose, necessity, and timing

## Travel
- Conference dissemination
- Collaboration meetings

## Materials and Supplies
- Consumables / software licenses

## Other Direct Costs
- Publication fees / participant incentives / consultants

## Subawards (if any)
- Scope and deliverables per partner

## Indirect Costs (F&A)
- Rate and base per institutional policy

Implementation Details

1) Agency-specific narrative mapping (what to write, where, and why)

• NSF

  • Two equal pillars: Intellectual Merit and Broader Impacts
  • Typical narrative pattern: problem → gap → approach → feasibility → outcomes → impacts
  • Ensure Broader Impacts are specific, measurable, resourced, and scheduled (not “bolt-on”).

• NIH

  • Core scored criteria: Significance, Investigator(s), Innovation, Approach, Environment
  • The Specific Aims page is the highest-leverage page: 2–4 aims, independent-but-complementary, each feasible with contingencies.
  • Approach must explicitly address rigor, reproducibility, and risk mitigation.

• DOE

  • Criteria vary by office; common expectations:
    • technical merit, mission relevance, team capability, facilities, and budget reasonableness
  • Often values integration of computation + experiment, partnerships, and (sometimes) cost share.

• DARPA

  • Emphasize: transformative payoff, measurable milestones, and transition.
  • Use phased plans with deliverables, metrics, and go/no-go criteria.
  • Answer DARPA-style questions in substance:
    • What if it works? Who cares? How will it transition?

• NSTC (Taiwan)

  • CM03 is central; feasibility and preliminary evidence are critical.
  • Plan for bilingual abstracts and include a clear research architecture diagram.

2) Visual requirement (mandatory minimum)

• Include at least 1–2 diagrams:
  • Workflow/method schematic (reduces reviewer cognitive load)
  • Timeline/Gantt with milestones and decision points
• Use consistent labeling, readable fonts, and captions that allow the figure to stand alone.

3) Milestones and risk control parameters

• Define milestones that are:
  • Measurable (metric + threshold)
  • Time-bound (quarter/year)
  • Decision-linked (go/no-go or pivot criteria)
• For each major risk, include:
  • failure mode → detection signal → mitigation → fallback method

4) Budget-to-workplan consistency checks

• Every major task should map to:
  • named personnel effort
  • required equipment/supplies
  • travel (if collaboration/fieldwork is claimed)
  • subaward scope (if partners are essential)
• Common rejection trigger: a narrative that promises outcomes without resourcing them in the budget.

5) Resubmission mechanics (especially NIH A1)

• Create a 1-page Introduction to Resubmission that:
  • lists major critiques
  • states exactly what changed and where
  • remains factual and respectful
• Strengthen the weakest scored criterion first (often Approach or Innovation), then tighten alignment across aims, methods, and milestones.