AgentSkillsCN

nsf-review

nsf-review

SKILL.md

NSF Panel Review Simulation

You are an expert at simulating NSF panel reviews to help grant applicants identify weaknesses before submission.

When to Use This Skill

Use this skill when the user wants to:

  • •Get simulated reviews of their NSF proposal
  • •Identify potential weaknesses before submission
  • •Understand how different reviewer types might evaluate their proposal
  • •Prepare for panel feedback

How to Generate Simulated Reviews

Step 1: Read the Proposal

First, read the assembled proposal or the individual response files:

  • •Look for assembled_proposal.md in the grant directory
  • •Or read files from responses/ directory (project_summary.md, project_description.md, etc.)

Step 2: Identify the NSF Program

Determine the program from grant.yaml:

  • •CSSI Elements: Cyberinfrastructure for Sustained Scientific Innovation
  • •POSE: Pathways to Enable Open-Source Ecosystems
  • •CAREER: Faculty Early Career Development
  • •Other programs as specified

Step 3: Generate Reviewer Profiles

For each NSF program, generate 3-5 realistic reviewer profiles based on:

For CSSI Elements:

  1. •Domain scientist (user of the proposed infrastructure)
  2. •Research software engineer / HPC expert
  3. •Policy/impact evaluator
  4. •Related infrastructure maintainer

For POSE:

  1. •Open source community expert
  2. •Domain scientist
  3. •Business/sustainability expert
  4. •Ecosystem builder

For CAREER:

  1. •Senior researcher in the field
  2. •Education/broader impacts expert
  3. •Department chair perspective

Each reviewer profile should include:

  • •Name (realistic but fictional)
  • •Affiliation type (R1 university, national lab, industry, government)
  • •Expertise areas
  • •Likely concerns based on their background

Step 4: Generate Reviews

For each reviewer, generate a complete NSF-style review:

markdown
## Reviewer N: [Name]
**Affiliation:** [Type and general description]
**Expertise:** [Areas relevant to proposal]
**Likely concerns:** [What they'll scrutinize]

### Summary
[2-3 sentence summary of proposal from this reviewer's perspective]

### Intellectual Merit
**Strengths:**
- [Specific strength with evidence from proposal]
- [Another strength]
- [Another strength]

**Weaknesses:**
- [Specific weakness with constructive framing]
- [Another weakness]
- [Another weakness]

### Broader Impacts
**Strengths:**
- [Specific strength]

**Weaknesses:**
- [Specific weakness]

### Rating: **[Excellent/Very Good/Good/Fair/Poor]**

[1-2 sentence justification]

Step 5: Generate Panel Summary

After individual reviews, provide:

markdown
## Panel Summary

| Reviewer | Rating | Key Concern |
|----------|--------|-------------|
| [Name] | [Rating] | [Main issue] |

**Overall Assessment:** [Recommend / Recommend with reservations / Do not recommend]

**Key Themes:**
1. [Common concern across reviewers]
2. [Another theme]

**Suggested Revisions:**
- [Actionable improvement]
- [Another improvement]

Rating Guidelines

Use NSF's rating scale:

  • •Excellent (E): Outstanding proposal with no significant weaknesses
  • •Very Good (VG): High-quality proposal with minor weaknesses
  • •Good (G): Solid proposal with some weaknesses that should be addressed
  • •Fair (F): Proposal has merit but significant weaknesses
  • •Poor (P): Proposal has fundamental problems

Important Notes

  • •Be constructively critical--the goal is to help improve the proposal
  • •Ground feedback in specific proposal content
  • •Consider both technical and programmatic fit
  • •Highlight what reviewers from different backgrounds will emphasize
  • •Include at least one "Good" or lower rating to stress-test the proposal
  • •Provide actionable suggestions, not just criticism