# MatchAwards [Health: Active]

**Category:** 🔎 Search & Data Extraction  
**Repository:** https://github.com/matchawards/matchawards-mcp  
**GitHub Stars:** 0  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/matchawards

## Description
Search US government contracts, grants and jobs from matchawards.com. Read-only, no key.

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `uvx` (confidence: high):

```json
"mcpServers": {
  "matchawards": {
    "command": "uvx",
    "args": ["matchawards-mcp"]
  }
}
```

## Documentation

## What MatchAwards MCP server does

MatchAwards MCP server exposes MatchAwards data to AI applications through the Model Context Protocol. It belongs in the 🔎 Search & Data Extraction category and is intended for read-only research into public procurement and funding data. No account or API key is required.

The catalog covers federal and state contract opportunities, federal grants, recent job postings, past federal contract awards, and grant awards. Results include structured fields and links back to MatchAwards; federal notices also include SAM.gov links when available. The tools do not modify records or submit applications, bids, or grants.

## How it works

An MCP client sends a tool request with filters such as keywords, NAICS codes, state, agency, set-aside, posting age, awardee, or award date. Search responses support one to 50 results and pagination, while NAICS lookup supports one to 25 codes. The server returns structured content with an output schema and also supplies the same JSON as text for clients that only read text.

Agency filtering uses a name lookup. Exact names are preferred, followed by names beginning with the requested text and then other API matches. The response identifies the selected agency and can list other matches. If no agency matches, the search covers all agencies and says so. Agency lookups are cached for 10 minutes.

## Setup and configuration

MatchAwards MCP server can be added as a remote MCP server using the hosted endpoint `https://matchawards.com/mcp`. The supplied material names Claude, Cursor, and VS Code as compatible MCP clients. The service is configured for no authentication, so running its tools does not require an account or API key.

Optional result cards are disabled by default. Setting `MATCHAWARDS_WIDGETS=1` enables the `show_results` tool and its inline MCP App resource for hosts that render MCP Apps. The hosted service also documents `MATCHAWARDS_USAGE_RETENTION_DAYS` for controlling usage-stat retention; its default is 180 days.

## Tools and capabilities

- `search_contracts` finds federal and state opportunities with filters for NAICS, state, keywords, set-asides, agency, posting age, and open status.
- `search_grants` searches federal grant and funding notices by NAICS, keyword, posting age, and open status.
- `search_jobs` searches job postings from the last 30 days using NAICS, state, keyword, and posting-age filters.
- `get_opportunity` returns a full opportunity record, published contacts, and any available AI-labelled notes.
- `lookup_naics` maps words or two- to six-digit code prefixes to NAICS codes.
- `search_awards` and `get_company` provide past award data and UEI-based company histories.
- `assess_opportunity` summarizes deadline status, qualification matches, similar awards, and winning firms. `find_teaming_partners` identifies firms with similar award histories.

## Limitations and notes

The MatchAwards MCP server is read-only and operates on the MatchAwards catalog. Award records can be incomplete, especially for federal awards. The assessment and teaming tools describe firms as having won similar work, not as incumbents. They use bounded opportunity and award requests, return the records they could read, and include warnings when limits stop the lookup early.

Company records provide award history but no contact details. Teaming results are based on public award data and may include firms without a UEI, which cannot be looked up further. Grant award titles use the recipient name, with the CFDA number used to distinguish the record. AI-generated notes are labelled as AI-generated and should be reviewed against the underlying opportunity data.

_Full upstream README: https://allmcps.com/mcp/matchawards/readme_