Grant Proposal Review

2. Ingest the proposal document

Handle the proposal based on its format:

Markdown or LaTeX: Read directly; no conversion needed.

PDF: Use a two-track approach:

1. Text extraction -- Read the PDF for content review. Use the Read tool directly on the PDF (Claude can read PDFs natively). For large PDFs (>10 pages), read in page ranges (e.g., pages: "1-10", then pages: "11-20"). For math-heavy or complex-layout PDFs where native reading struggles, convert to markdown via opencite:

   uvx opencite convert proposal.pdf -o proposal.md
   

2. Visual layout analysis -- Convert each page to PNG for figure sizing and space utilization review:

   uv run --with pdf2image --with pillow python -c "
   from pdf2image import convert_from_path
   pages = convert_from_path('proposal.pdf', dpi=150)
   for i, page in enumerate(pages):
       page.save(f'proposal_page_{i+1}.png', 'PNG')
   "
   

   Note: pdf2image requires poppler as a system dependency (brew install poppler on macOS, apt install poppler-utils on Linux). If poppler is not available, use pdftoppm -png -r 150 proposal.pdf proposal_page directly, or fall back to reading the PDF natively with the Read tool.

   Read each page image to assess:

   • Are figures appropriately sized for their content, or oversized with wasted space?
   • Are there large areas of whitespace or underutilized regions?
   • Could any figures be reduced without losing clarity?
   • Are margins and spacing consistent throughout?
   • Is text density appropriate (not too sparse, not too cramped)?

   Include space utilization observations in the review output under a "Layout and Space Utilization" section.

Read all submitted sections. For NIH proposals, this typically includes:

• Specific Aims (1 page)
• Research Strategy: Significance, Innovation, Approach
• Any supporting materials (biosketch, facilities, data management plan)

For NSF proposals:

• Project Summary
• Project Description
• Data Management Plan

3. Score each criterion

NIH Scoring (1-9 scale):

For each of the five review criteria, assign a score and provide justification:

Score descriptors:

• 1 (Exceptional): Essentially no weaknesses
• 2 (Outstanding): Extremely minor weaknesses
• 3 (Excellent): Minor weaknesses
• 4 (Very Good): Some moderate weaknesses
• 5 (Good): Moderate weaknesses
• 6 (Satisfactory): Some major weaknesses
• 7 (Fair): Major weaknesses
• 8 (Marginal): Numerous major weaknesses
• 9 (Poor): Fundamentally flawed

NSF Rating:

• Excellent / Very Good / Good / Fair / Poor
• Evaluate Intellectual Merit and Broader Impacts separately

4. Identify overall strengths and weaknesses

Synthesize across criteria. Focus on:

• Strengths: What makes this proposal competitive?
• Weaknesses: What would a skeptical reviewer flag?
• Fatal flaws: Issues that would prevent funding regardless of other merits

5. Produce the review output

Structure the output as described below.

Review Output Format

Structure the review output according to the appropriate agency template in references/review-output-templates.md. Both NIH and NSF templates follow this general structure:

1. Summary - 2-3 sentence proposal overview
2. Criterion scores - Individual scores (NIH 1-9) or ratings (NSF Excellent-Poor) with strengths/weaknesses for each criterion
3. Additional review criteria - Non-scored items (human subjects, data management, rigor)
4. Layout and space utilization (when PDF was provided) - Observations on figure sizing, whitespace usage, areas where space could be saved or better utilized, and whether the proposal makes effective use of its page limits
5. Actionable improvements - Prioritized as Critical, Important, and Suggested

For a complete worked example, see examples/sample-nih-r01-review.md.

Review Perspective

Adopt the viewpoint of a senior researcher on a study section or review panel:

• Expertise: Assume deep domain knowledge; do not flag common techniques as novel
• Skepticism: Demand evidence for claims; flag unsupported assertions
• Constructiveness: Every weakness should include a suggestion for improvement
• Fairness: Acknowledge strengths genuinely; do not manufacture weaknesses
• Calibration: Score relative to the mechanism (R21 should not be held to R01 preliminary data standards; K awards emphasize career development over research scope; CAREER proposals require genuine research-education integration; DP2 rewards bold, innovative thinking from new investigators)
• Precision: Cite specific sections, figures, or claims when identifying issues
• Impact focus: Always tie feedback back to how it affects the overall impact score

Common Issues

For common reviewer comments and their meanings, consult references/review-best-practices.md.

Additional Resources

Reference Files

• references/nih-review-criteria.md - Complete NIH review criteria, scoring rubric, and study section process
• references/nih-career-training-criteria.md - Review criteria for K, F, and T32 mechanisms
• references/nsf-review-criteria.md - Complete NSF review criteria and panel process
• references/review-best-practices.md - Best practices from experienced reviewers, common reviewer comments, and calibration guidance
• references/review-output-templates.md - NIH and NSF review output format templates

Examples

• examples/sample-nih-r01-review.md - Complete example review demonstrating expected format, tone, and scoring calibration