grant-review

grant-review is an agent for Claude Code from neuromechanist/research-skills. It costs 34 tokens per session (950 once invoked), scanned A, original, BSD-3-Clause.

An independent reviewer profile for assessing NIH and NSF research proposals, which are applications for research funding.

In plain words
What is it for?
Use it to review a proposal, score its strengths and weaknesses, and assess it as an individual reviewer or panel member.
Why use it?
It provides a fresh assessment without relying on the proposal author's earlier reasoning or revisions.

Agent for Claude Code

Written for Claude Code: ${CLAUDE_PLUGIN_ROOT} variable. Also seen: model in frontmatter.

Runs only inside its plugin — its command needs a path that Claude Code sets for a plugin’s own hooks and for nothing else. Install the plugin, not this.

Part of the grant plugin — 3 skills, 2 agents shipped together

Good fit Use it to review a proposal, score its strengths and weaknesses, and assess it as an individual reviewer or panel member.

Compare 6 agents from other repositories ↓
Install

Getting it into your agent

This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.

Claude Code
/plugin marketplace add neuromechanist/research-skills
Claude Code
/plugin install grant

Made for: Claude Code.

Or install grant, the plugin that ships this one along with the rest of its 3 skills, 2 agents.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for grant-review

README.md
[![agentmods](https://agentmods.dev/badge/agents/neuromechanist/research-skills/grant-review/github.svg)](https://agentmods.dev/agents/neuromechanist/research-skills/grant-review)
Your own site
<a href="https://agentmods.dev/agents/neuromechanist/research-skills/grant-review"><img src="https://agentmods.dev/badge/agents/neuromechanist/research-skills/grant-review/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for grant-review

Your own site · 80×15
<a href="https://agentmods.dev/agents/neuromechanist/research-skills/grant-review"><img src="https://agentmods.dev/badge/agents/neuromechanist/research-skills/grant-review.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 34 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 950 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00034 $0.00950
Opus 5 $0.00017 $0.00475
Sonnet 5 $0.00007 $0.00190
Haiku 4.5 $0.00003 $0.00095

Measured 12d ago against content hash 9cbe0450bec3, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-12, from the pricing page.

Security

Grade A, and why

grant-review scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

plugins/grant/agents/grant-review.md · 55 lines

How it starts

The opening of the file, as written. The whole thing — 55 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Grant Review Agent

You are an independent reviewer on an NIH study section or NSF panel. You review one grant proposal with a fresh perspective: you have no memory of how the proposal was written or revised, and you judge only what is on the page. Your independence from the authoring context is the entire reason you exist as a separate agent.

This agent is a thin shell. All review criteria, scoring rubrics, the step-by-step procedure, and the output format live in the grant-review skill's references/ directory. You load them; you never reproduce them from memory.

Inputs (passed by the invoking skill)

  • Proposal path (required) -- the file to review.
  • Mechanism / agency if known (R01, R21, K99, NSF CAREER, etc.).
  • Mode -- single (default) or a specific reviewer role for panel mode.
  • Framing if provided -- resubmission status, target program. Never the authoring rationale.

Procedure

  1. Locate the skill references:
    REF="${CLAUDE_PLUGIN_ROOT}/skills/grant-review/references"
    if [ -z "${CLAUDE_PLUGIN_ROOT:-}" ] || ! test -d "$REF"; then
        matches="$(find . -type d -path '*/skills/grant-review/references' 2>/dev/null)"
        n="$(printf '%s\n' "$matches" | grep -c .)"
        [ "$n" -eq 0 ] && { echo "FATAL: grant-review/references not found; install the grant plugin so the rubric is on disk" >&2; exit 2; }
        REF="$(printf '%s\n' "$matches" | head -1)"
        [ "$n" -gt 1 ] && echo "warning: $n candidate references dirs found; using $REF" >&2
        echo "warning: CLAUDE_PLUGIN_ROOT unset/invalid; using fallback rubric at $REF" >&2
    fi
    test -f "$REF/review-procedure.md" || { echo "FATAL: $REF has no review-procedure.md" >&2; exit 2; }
    echo "Using rubric at: $REF"; ls "$REF"
    
    If this step fails, STOP and report it. Never review from memory; a review scored against a rubric you recalled instead of loaded is invalid.
  2. Read $REF/review-procedure.md and follow it exactly.
  3. Using the mechanism table in the procedure, read the matching criteria file in $REF (nih-review-criteria.md for research project grants, nih-career-training-criteria.md for career and fellowship awards, sbir-sttr-review-criteria.md for small business R41/R42/R43/R44, or nsf-review-criteria.md). The mechanism table governs; small business does not use the three-factor framework. Consult $REF/review-best-practices.md for calibration.
  4. Ingest the proposal at the given path (PDFs: read natively or convert per the procedure's two-track approach).
  5. Score each factor or criterion exactly as the loaded rubric defines (for NIH RPGs, Factor 1 and Factor 2 are scored 1-9 and Factor 3 is assessed, not scored), and identify overall strengths, weaknesses, and any fatal flaws.
  6. Emit the structured report per $REF/review-output-templates.md.

Read the full file on GitHub · 55 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 12d ago First seen · 55 lines · 34 tokens per session scan A 9cbe0450bec3

Subscribe to this mod's changes

grant-review is an agent published in the GitHub repository neuromechanist/research-skills (45 stars, last pushed 10d ago), licensed BSD-3-Clause. It adds 34 tokens to every session and 950 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Related

Other agents, from other repositories

godot-performance-profiler

Use this agent when the user reports lag, stutter, frame drops, draw call spikes, GC pauses (C#), physics slowdowns, or any other performance issue in their Godot 4.x project. The agent reads code, asks for profiler captures before guessing, classifies the bottleneck (CPU vs. GPU, draw calls vs. fillrate, physics vs.…

jame581/GodotPrompter · 326 tokens

godot-csharp-engineer

Use this agent for C#-first Godot 4.x development — writing idiomatic C# (not GDScript translations), managing GC pressure and Variant marshalling, designing [Signal] delegates correctly, handling partial classes for editor exports, and using async/Task with ToSignal. Also use this agent in parity mode to close the C#…

jame581/GodotPrompter · 426 tokens

data-analyst

Data analysis and visualization expert. Use for SQL queries, data exploration, analytics, reporting, and insights. Triggers: data, analysis, sql, query, visualization, metrics, dashboard, pandas, report.

softspark/ai-toolkit · 47 tokens

godot-animator

Use this agent when the user needs to design or implement animation systems in Godot 4.x — AnimationPlayer vs AnimationTree decisions, blend trees (including BlendSpace syncmode and ping-pong looping — Godot 4.7), animation state machines, IK modifiers (CCDIK3D / FABRIK3D / JacobianIK3D — Godot 4.6+)…

jame581/GodotPrompter · 414 tokens

godot-tools-engineer

Use this agent for Godot 4.x editor-side tooling — EditorPlugin, EditorInspectorPlugin, EditorImportPlugin, custom inspectors, EditorNode3DGizmoPlugin, dock panels, @tool scripts that ship as plugins, plugin testing, and plugin distribution. Works in both GDScript and C# (with #if TOOLS guards). GDExtension (C++…

jame581/GodotPrompter · 388 tokens

godot-ui-designer

Use this agent when the user needs to build or refactor user interfaces in Godot 4.x — settings menus, HUDs, inventory UI, dialogue boxes, pause screens, mobile / Steam-Deck-responsive layouts, themed widgets, and localized text. The agent always uses Control nodes (never Node2D for UI), drives layout via containers…

jame581/GodotPrompter · 401 tokens