AGENT SKILL

gp-potential-reviewers

Compile the list of people likely to review or sit on the panel for an NSF ANT (Algebra and Number Theory) proposal — the PI's suggested reviewers, recent ANT awardees from the nsf.gov award search (program element 1264, last two years), authors the proposal cites heavily, and senior people in the subfield — minus the proposal's COA conflicts, written to submission-docs/potential-reviewers.md. Use for "who will review this", "list potential reviewers", "who won ANT last year", "target the panel", or before pd-review-as runs its perspective passes. Never guesses emails; never contacts anyone.

License 许可
GPL-3.0
最近更新
2026-11-13
Stars
★ 0
Forks
0

01 前言 Front Matter

name
gp-potential-reviewers
description
Compile the list of people likely to review or sit on the panel for an NSF ANT (Algebra and Number Theory) proposal — the PI's suggested reviewers, recent ANT awardees from the nsf.gov award search (program element 1264, last two years), authors the proposal cites heavily, and senior people in the subfield — minus the proposal's COA conflicts, written to submission-docs/potential-reviewers.md. Use for "who will review this", "list potential reviewers", "who won ANT last year", "target the panel", or before pd-review-as runs its perspective passes. Never guesses emails; never contacts anyone.

03 SKILL.md 正文

gp-potential-reviewers

Produces the perspective list for pd-review-as and the fairness list for PD-POLICY P14 (every senior person whose work the draft uses is named and cited). It is a reading aid, not a submission document: NSF receives only reviewers.txt. No one on the list is contacted.

Inputs

  • The proposal folder: proposal/tex/description.tex, references.bib, submission-docs/reviewers.txt, submission-docs/COA.xlsx.
  • The program element (ANT is 1264) and the two most recent fiscal years.

Procedure

  1. PI's suggested reviewers. Read reviewers.txt; every name goes in, marked suggested.

  2. Recent awardees. Query the NSF award search (nsf.gov/awardsearch, advanced search, program element code 1264, start date within the last two fiscal years; the export is CSV). Record PI name, institution, title, award number, year. Take all; mark awardee. Record the query date and the count. If the site cannot be reached, say so and leave the section marked not run; do not fill it from memory.

  3. Heavily cited authors. From references.bib and the \cite counts in description.tex, list authors cited twice or more or whose theorem a project rests on; mark cited. Program officers for ANT (from the program page, dated) go in marked program officer.

  4. Senior people in the subfield. From the cited authors' coauthor networks and the PI's own knowledge of the field, the people whose problems the proposal attacks or whose results it uses (for differential algebra: the model theorists of differential fields, the Galois-theory and D-module people, the computational differential algebra people). Mark senior. For each, one line on which part of the draft touches their work; this is the P14 fairness check input.

  5. Exclude conflicts. Open COA.xlsx read-only (gp-coa describes the tables): remove anyone in Table 2 (collaborators) or Table 4 (advisors and advisees), and anyone at the PI's institution; keep them in a separate "conflicted, excluded" list so the exclusion is visible.

  6. Write submission-docs/potential-reviewers.md, dated:

    | name | institution | subfield | source | why they matter to this draft | likely panelist? |

    source is one of suggested, awardee <year>, cited (n), senior, program officer. "Likely panelist" is yes for suggested reviewers, last year's awardees in the same subfield, and senior people the draft leans on; pd-review-as runs one pass per yes. Then the conflicted list, then the queries run with dates and counts.

  7. Report counts per source and the yes list. Do not edit reviewers.txt; the PI chooses suggested reviewers.

Red flags

Thought Reality
"I remember who won ANT in 2025" Run the award search; record date and count.
"Their address is probably name at their institution" No emails; names and institutions only.
"Add them to reviewers.txt too" The PI chooses suggested reviewers (gp-researchgov).
"A collaborator would review it fairly" COA Table 2 and 4 names are excluded, always.
"Draft a note to ask if they would review" No contact of any kind.

04 同仓库更多 Skills