nsf-awards-mcp
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@nsf-awards-mcpFind NSF awards related to artificial intelligence"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
nsf-awards-mcp
MCP server for NSF award and funding data, packaged for one-line activation.
npx -y github:MCP-Hive/nsf-awards-mcpThis is a fork of @access-mcp/nsf-awards
by ACCESS-CI, patched so that it starts when launched through npx. All of the
NSF query logic is upstream's; see Relationship to upstream.
What it does
One tool, search_nsf_awards, over the NSF Award Search API:
Parameter | Type | Purpose |
| string | Fetch a specific award by number, e.g. |
| string | Search keywords in titles and abstracts |
| string | Look up awards by principal investigator |
| string | Filter by awardee organization |
| boolean | Only awards where the institution is the primary recipient |
| number | Max results (default 10) |
Returns {total, items}, where each item carries award number, title,
institution, PI and co-PIs, intended and to-date amounts, start and end dates,
and the abstract.
The upstream NSF API needs no API key, no registration, and no payment. NSF award data, abstracts included, is a US government work in the public domain.
Related MCP server: crossref-mcp
Why this fork exists
Upstream's src/index.ts ends with:
if (import.meta.url === `file://${process.argv[1]}`) {
main().catch(...);
}npx runs a package through its bin symlink. Node resolves symlinks when
computing import.meta.url but leaves process.argv[1] unresolved, so the two
are never equal, main() is never called, and the process exits 0 without ever
speaking JSON-RPC. The client sees only a closed connection:
$ npx -y @access-mcp/nsf-awards
(exits 0, no output)This fork calls main() unconditionally, matching what the other servers in the
upstream monorepo already do.
Reported upstream as necyberteam/access-mcp#66. If that is fixed and released, this fork stops being necessary.
Relationship to upstream
Forked from necyberteam/access-mcp at commit
99004ce.
Changes made here, and nothing else:
src/index.ts— removed the main-module guard described above.vendor/access-mcp-shared/— vendoredpackages/sharedfrom the same commit. The published@access-mcp/shared@0.9.0tarball is older than its own source tree and does not exportprojectFields, whichsrc/server.tsimports, so resolving that dependency from npm does not build. Vendoring at the pinned commit keeps the tree self-consistent.Packaging — a single bundled
server/bundle.mjsand a rootbin, so thenpx -y github:form works.
src/server.ts is unmodified.
Packaging notes
Upstream is a workspaces monorepo whose root has no bin, so npx against a
plain fork of it has nothing to run. This repo is therefore a standalone
extraction with one bin at the root.
build-bundle.mjs inlines the source and every runtime dependency into
server/bundle.mjs, which is committed. The package declares no runtime
dependencies and no prepare script, so npx -y github:... clones the repo
and immediately executes — nothing to install, nothing to compile. That matters
when the server is cold-started inside a Lambda container.
The bundle is ESM because src/server.ts uses createRequire(import.meta.url).
A createRequire shim is injected after the shebang for the CommonJS transitive
dependencies that call require() for node builtins.
Development
npm install # dev dependencies only, needed to rebuild
npm run bundle # rebuild server/bundle.mjs
npm run smoketest # spawn the bundle, list tools, run a live NSF queryVerify the published one-liner end to end:
npm run smoketest -- github:MCP-Hive/nsf-awards-mcpCommit the rebuilt server/bundle.mjs — it is the artifact npx executes.
License
MIT, inherited from ACCESS-CI. See LICENSE.
Available Tools
1 toolsearch_nsf_awardsB
Search NSF awards and funding. Returns {total, items}.
| Name | Required | Description | Default |
|---|---|---|---|
| id | No | Award number (e.g., '2138259') | |
| pi | No | Principal investigator name | |
| limit | No | Max results (default: 10) | |
| query | No | Search keywords in titles/abstracts | |
| fields | No | Project the response down to only these fields. Dotted path syntax: 'total', 'items[].title', 'items[].principalInvestigator', 'metadata.pagination.has_more', etc. Use to reduce payload size when you only need specific fields. Omit to receive the full response. Only applies to listing results, not single-award lookups. | |
| institution | No | Institution name | |
| primary_only | No | When searching by institution, only return awards where the institution is the PRIMARY recipient (excludes collaborative/co-PI awards from other institutions). Default: false |
TDQS
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations are provided, so the description carries the full behavioral burden. It only discloses the top-level return shape and does not address pagination, rate limits, authentication requirements, or read-only behavior.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is a single, front-loaded sentence with no filler. It communicates the resource and the top-level output shape in under ten words.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
The schema fully documents all seven optional parameters, making the tool callable in principle. However, the absence of an output schema and the sparse description leave gaps around item structure, pagination, and how multiple filters interact.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 100%, and all parameters have detailed help text in the schema, so the baseline of 3 applies. The tool description itself adds no parameter-level meaning beyond the generic 'search' context.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description identifies a specific verb ('Search') and target resource ('NSF awards and funding') and states the top-level return shape. There are no sibling tools to differentiate from, so the lack of explicit comparison is not penalized.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description implies the tool is used for searching NSF award data, but it does not state when to use it versus alternatives or mention any exclusions. With no sibling tools or additional context, the usage guidance is only implicit.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections. Dates show when Glama detected each change.
1 tool update
v0.3.1- First observed
search_nsf_awards
TDQS
With only one tool, there is no possibility of confusion between tools. The tool's purpose is clearly described as searching NSF awards.
The single tool name 'search_nsf_awards' follows a clear verb_noun snake_case convention, which is consistent with standard MCP naming practices and predictable.
At one tool, the server feels thin, but it is scoped to a specific search function. This is borderline for the stated purpose of searching NSF awards and funding.
The search tool covers the primary need, but there are minor gaps such as lacking a direct 'get award by ID' endpoint. Agents can work around this by searching for and retrieving specific items from the results.
Maintenance
Resources
Unclaimed servers have limited discoverability.
Looking for Admin?
If you are the server author, to access and configure the admin panel.
Related MCP Connectors
NSF Awards MCP — National Science Foundation award database (free, no auth)
Access US federal award, recipient, agency, and spending analytics data from USAspending.gov.
Search US grants + federal contracts (Grants.gov + SAM.gov) from any LLM.
SBIR MCP — wraps the SBIR.gov public API (free, no auth)
Related MCP Servers
- FlicenseNot gradedqualityDmaintenanceProvides programmatic access to the NIH Reporter API for searching and retrieving detailed information on NIH-funded research projects, grants, and investigators. It enables AI assistants to query project abstracts, funding amounts, and organization details through natural language.-
- FlicenseAqualityBmaintenanceEnables searching and retrieving scholarly metadata from Crossref's public REST API, including works, journals, funders, and members, without requiring an API key.101-
- AlicenseNot gradedqualityCmaintenanceAccess data on every NIH-funded research project. Free and no authentication required.6MIT
- AlicenseNot gradedqualityCmaintenanceEnables querying the National Science Foundation award database without authentication.5MIT
Latest Blog Posts
- Who's Calling? MCP Hosts Are an Identity Blind Spot (And the Spec Knows It)By Om-Shree-0709 on .mcpAgent IdentityOAuth 2.1
- Your AI Chatbot Just Exposed Your CEO's Salary to an InternBy Om-Shree-0709 on .Agent IdentityMCP SecurityOAuth Delegation
- Why MCP Servers Need Execution Sandboxing (And Why Your Current Stack Isn't Enough)By Om-Shree-0709 on .Agentic AiPrompt InjectionWebAssembly
MCP directory API
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/MCP-Hive/nsf-awards-mcp'
If you have feedback or need assistance with the MCP directory API, please join our Discord server