Skip to main content
Glama
cliwant

mcp-sam-gov

by cliwant
OverviewSchemaSourceRelated ServersDiscussionsAdmin
TypeScript
Remote

nsf_search_awards

Read-only

Search NSF research-grant awards by keyword, state, UEI, principal investigator, date range, and more. Filter results with pagination to find grant data joinable to SAM.gov.

Instructions

Search awarded NSF research-grant awards (keyless; api.nsf.gov/services/v1/awards.json), joinable to SAM/USAspending via ueiNumber/parentUeiNumber. Filters: keyword (MULTI-WORD is OR-tokenized — 'machine learning' = machine OR learning, disclosed), awardeeStateCode (UPPERCASE 2-letter USPS — non-state typo silently returns 0), awardeeName, ueiNumber (12-char UEI — EXACT SAM/USAspending join), parentUeiNumber, pdPIName, dateStart/dateEnd (strict mm/dd/yyyy — wrong format silently mis-parsed), limit (1..100), offset (0..9999). Returns { awards:[{ id, title, agency, cfdaNumber, transType, awardee:{name, city, stateCode, ueiNumber, parentUeiNumber}, principalInvestigator, coPrincipalInvestigators, programOfficer, amounts:{fundsObligatedAmt, estimatedTotalAmt, fundsObligatedByYear}, dates, program, activeAward, historicalAward }] } (abstract EXCLUDED — use nsf_get_award) + honest _meta. HONESTY: NSF Awards are RESEARCH GRANTS, NOT procurement contracts (ueiNumber joins SAM/USAspending but the award nature differs — disclosed every response). totalAvailable = EXACT metadata.totalCount below 10,000; SATURATES at 10,000 (ES track_total_hits cap → totalIsLowerBound:true + note; first 10,000 only retrievable). NSF caps retrieval at offset+rpp ≤ 10,000 (offset ≥ 10,000 → invalid_input). fundsObligatedAmt/estimatedTotalAmt: STRINGS → number|null (genuine $0 is 0, absent is null). Genuine totalCount:0 → complete:true/total:0; serviceNotification at HTTP 200 → THROWS; outage/5xx THROWS; 200 not {response:{award,metadata}} or non-numeric totalCount → schema_drift.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
limitNoAwards per page (→ NSF rpp), 1..100, default 25. The OUTGOING page size is clamped so offset+rpp ≤ 10,000 (crossing NSF's retrieval window triggers a FATAL).
offsetNo0-based offset. HARD-CAPPED at 9,999: NSF caps keyless retrieval at the first 10,000 records (offset+rpp ≤ 10,000), so offset ≥ 10,000 is refused (invalid_input) — narrow criteria to bring the set under 10,000.
dateEndNoAward ACTION-date upper bound. STRICT mm/dd/yyyy (same semantics/foot-gun as dateStart). e.g. '12/31/2024'.
keywordNoFree-text search over title/abstract. NOTE: NSF OR-tokenizes a MULTI-WORD keyword (matches ANY word, not the phrase — 'machine learning' = machine OR learning, a far broader set; disclosed in _meta.notes). Use a single distinctive word or add a scoping filter for a precise set.
pdPINameNoPrincipal-investigator name filter (2..120 chars). LIVE-CONFIRMED to narrow. e.g. 'Bell'.
dateStartNoAward ACTION-date lower bound (the initial award/obligation date, NOT the project startDate — live-verified). STRICT mm/dd/yyyy; a wrong format (yyyy-mm-dd) is silently mis-parsed by NSF (not an error), so it is rejected. e.g. '01/01/2024'.
ueiNumberNoAwardee UEI — a 12-char alphanumeric SAM/USAspending Unique Entity ID (uppercase-normalized before sending). LIVE-CONFIRMED an EXACT recipient-graph filter (the clean SAM/USAspending join). e.g. 'FTMTDMBR29C7' (Johns Hopkins).
awardeeNameNoAwardee-organization name filter (2..200 chars). LIVE-CONFIRMED to narrow (a top recipient like 'Johns Hopkins University' may still saturate at the 10,000 count cap).
parentUeiNumberNoParent-organization UEI — a 12-char alphanumeric UEI for the awardee's parent entity (uppercase-normalized). LIVE-CONFIRMED an EXACT narrow (the parent-org roll-up join). e.g. 'GS4PNKTRNKL3'.
awardeeStateCodeNoAwardee-organization US state/territory 2-letter USPS code (UPPERCASE — the enum is the SSRF value guard + the silent-zero guard: a non-state typo silently returns 0 awards on NSF, indistinguishable from 'no NSF funding', so it is an invalid_input). LIVE-CONFIRMED to narrow. e.g. 'CA'.

Schema Changelog

Changes observed during successful MCP inspections.

  1. Addedv1.12.0

TDQS

A4.7/5.0
Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

With readOnlyHint and openWorldHint annotations already covering safety and openness, the description goes far beyond them: it discloses silent date mis-parsing, OR-tokenization of multi-word keywords, saturation at 10,000 with totalIsLowerBound, offset hard-cap, string-to-number coercion, error throwing on serviceNotification/5xx, and schema_drift detection. This is the model of behavioral transparency.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Although long, the description is tightly organized into Filters, Returns, and HONESTY sections with every sentence carrying operational value. The most critical behavioral caveats are front-loaded and scannable, making the length appropriate for a 10-parameter tool with multiple silent-failure footguns.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

The description is complete for invoking and interpreting results: it specifies the return shape, field semantics, saturation behavior, join semantics, error conditions, and alternative-tool routing. Even without an output schema, an agent has everything needed to call correctly and interpret the response.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

The schema description coverage is 100% and the schema descriptions are already very rich, containing the OR-tokenization warning, strict date format, UEI exact-join semantics, and offset cap. The prose redistributes this information but adds little new meaning beyond the schema, so the baseline of 3 applies.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description names a specific verb plus resource ('Search awarded NSF research-grant awards'), adds the keyless API endpoint, and explicitly contrasts with nsf_get_award ('abstract EXCLUDED — use nsf_get_award'). It leaves no ambiguity about what this tool retrieves or how it differs from its sibling.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines5/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

It explicitly tells the agent when to use this tool versus the named alternative: use this for award search without abstracts, use nsf_get_award when abstracts are needed. It also gives practical usage constraints, such as narrowing criteria when retrieval caps at 10,000 records and using precise filters for a clean set.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Deploy Server

Other Tools