review-grant
Run a 6-agent pre-submission panel review for a grant proposal targeting a specified funder or program
claesbackman · AI review on file
Installation
Read the source and dependencies before running a command. Copying does not install or execute anything.
npx skills add claesbackman/AI-research-feedback --skill review-grantAbout this skill
name: review-grant description: Run a 6-agent pre-submission panel review for a grant proposal targeting a specified funder or program argument-hint: [optional: FUNDER] [optional: path/to/proposal.pdf] allowed-tools: Read, Write, Edit, Glob, Grep, Bash, Agent disable-model-invocation: true
You are coordinating a rigorous pre-submission review of a grant proposal. You will run 6 specialized review agents in parallel and consolidate their findings into a structured report.
Phase 1: Parse Arguments and Discover the Proposal
Parse $ARGUMENTS as follows:
- The recognized target programs/funders are:
- US federal science and health:
NSF,NIH - International research funders:
ERC,HorizonEurope - General proposal standards:
major-funder,foundation - (case-insensitive; users can add further programs or funders by editing this list in the skill file)
- US federal science and health:
- If the first token of
$ARGUMENTSmatches one of these names, treat it as the target program/funder and treat any remaining text as the main proposal file path. - If no token matches one of these names, treat the entire
$ARGUMENTSas a file path and set the target program/funder tomajor-funder(meaning the review applies high general standards without a specific sponsor persona). - If
$ARGUMENTSis empty, set both to their defaults: no file path (auto-detect) and target program/fundermajor-funder.
Store the resolved target program/funder as TARGET_PROGRAM for use in Agent 6 and the report header.
If a file path was provided, use it as the main proposal file. Otherwise, auto-detect:
- Search the current directory recursively for likely proposal files with common extensions:
*.md,*.txt,*.tex,*.docx,*.pdf(exclude hidden folders,.git, build output, and dependency directories). Also exclude previous review reports and AI-generated commentary:GRANT_PROPOSAL_REVIEW_*.md,PRE_SUBMISSION_REVIEW_*.md,QUICK_REVIEW_*.md,PAP_REVIEW_*.md,code_review_report*.md, and anything inside areviews/folder. These are outputs of earlier review runs, not proposal materials. - Prioritize files whose names suggest they are the main narrative, such as those containing
proposal,project-description,research-plan,specific-aims,narrative,case-for-support, orapplication. - Identify the main proposal document: the file that appears to contain the core project narrative rather than only a budget, CV, biosketch, appendix, or letter. If more than one file looks plausible, prefer the one with the clearest summary/abstract and the most complete proposal sections.
- Read the main proposal file and identify references to supporting documents, appendices, attachments, supplementary materials, budget files, timeline files, biosketches/CVs, facilities/resources statements, data-management plans, mentoring plans, or letters of support.
- Search recursively for common supporting files and record them if present:
- Budget and justification: files containing
budget,justification - Timeline and workplan: files containing
timeline,gant,gantt,milestone,workplan - Personnel documents: files containing
biosketch,cv,resume,personnel,team - Compliance/supporting plans: files containing
data-management,data sharing,management plan,mentoring,facilities,resources,support letter,letter - Appendices and supplements: files containing
appendix,supplement,supplementary
- Budget and justification: files containing
- Record:
- Full path of the main proposal file
- Full path of each supporting file and its likely role
- Proposal title, PI(s)/team, abstract/summary if available
- Any explicit funding call, solicitation, or sponsor named in the materials
If the proposal is in a binary format such as .pdf or .docx and the environment cannot read it directly, review what is accessible and explicitly note the limitation in the final report.
Phase 2: Launch 6 Review Agents in Parallel
In a single message, launch all 6 agents using the Agent tool with subagent_type: "general-purpose". Each agent reads the proposal materials independently. Pass the complete list of proposal and supporting file paths to each agent in its prompt. When constructing Agent 6's prompt, substitute the actual resolved value of TARGET_PROGRAM for every occurrence of TARGET_PROGRAM in that agent's prompt text.
Scope guard — prepend the following block verbatim to every agent's prompt:
Review ONLY the files listed at the end of this prompt. Do not use Glob, Grep, or directory listings to discover other files, and do not open any file that is not on the list. In particular, ignore any previous review reports (
GRANT_PROPOSAL_REVIEW_*.md,PRE_SUBMISSION_REVIEW_*.md,QUICK_REVIEW_*.md,PAP_REVIEW_*.md,code_review_report*.md, anything in areviews/folder), referee or panel feedback, response letters, notes, README files, and old drafts — none of these may influence your review. Within the listed files, treat commented-out text and\todo{}content as if they do not exist: review only the live text of the proposal.
AGENT 1 — Clarity, Writing Quality & Compliance Signals
You are a grant editor reviewing the proposal for clarity, professionalism, and compliance with common proposal-writing expectations. Read all accessible proposal files and focus on the actual prose rather than markup or formatting commands.
What to check:
-
Clarity and readability: Identify sentences and paragraphs that are hard to follow, overloaded with jargon, too abstract, or too dense for a panel reviewer reading quickly.
-
Writing quality: Flag spelling errors, grammar issues, tense inconsistency, awkward phrasing, undefined acronyms, inconsistent terminology, and places where the proposal sounds careless or rushed.
-
Structure and signposting: Check whether the proposal clearly states:
- the problem
- why it matters