ca-grants
Enables searching and retrieving California state grant opportunities, including filters by status, category, and agency, without requiring an API key.
README
@pipeworx/ca-grants
California Grants Portal MCP — open State of California grant funding opportunities: who administers them, who may apply, how much money is available, and when applications are due. Keyless.
Tools
ca_search_grants(query?, status?, category?, applicant_type?, limit?, offset?)— free-text search over California state grant opportunities, filterable by status, category, and eligible applicant type. Defaults tostatus="active".ca_get_grant(portal_id)— the full record for one grant: purpose, description, eligibility notes, geography, funding source, matching-funds requirement, award amounts, funding method, contact info, and links.ca_grants_closing_soon(days?, category?, applicant_type?, limit?)— active grants with a fixed deadline in the next N days (default 30), soonest first, plus a count of rollingOngoingopportunities.ca_grants_by_agency(agency, status?, limit?, offset?)— grants from one California department/agency, matched as a forgiving case-insensitive substring, with a status breakdown.ca_list_grant_categories(status?)— the category, applicant-type, agency, and status values actually present in the data, each with a grant count, so filter values can be read off real data.
Auth
Keyless. No API key, no BYO credentials.
Data sources
- Dataset:
https://data.ca.gov/dataset/california-grants-portal - Endpoint:
https://data.ca.gov/api/3/action/datastore_search - Resource id:
111c8c88-21f6-453c-ae2c-b4785a0624f5(~1,974 rows, refreshed daily) - Human portal:
https://www.grants.ca.gov/
Gotchas
Field ids are TitleCase (AgencyDept, ApplicationDeadline, EstAvailFunds) and getting the case wrong makes CKAN silently return nothing rather than erroring, so every field name in this pack is spelled from the live fields list. Every value is stored as text, which has three consequences. EstAvailFunds is a display string like "$750,000" and EstAmounts is often a range like "Between $1 and $375,000" — this pack deliberately exposes no numeric minimum-amount filter, because any such filter would have to guess at ranges; est_available_funds_approx parses the leading figure as advisory context only and is never filtered on. Categories and ApplicantType are semicolon-joined lists ("Nonprofit; Public Agency; Tribal Government"), so they are split per-atom and matched case-insensitively client-side rather than through CKAN filters, which only does equality/IN. And ApplicationDeadline is text: usually "YYYY-MM-DD HH:MM:SS", but 127 of the 177 currently-active rows hold the literal string "Ongoing" (rolling intake, no due date) with one row null — so deadline math happens client-side and ca_grants_closing_soon reports ongoing_no_deadline as its own bucket rather than dropping two-thirds of the open grants on the floor. Deadline timestamps carry no timezone and are treated as UTC, which is accurate to within a day. Status takes exactly three values (active 177, closed 1,796, forecasted 1), so unfiltered searches are dominated by expired opportunities and ca_search_grants defaults to active. Because CKAN cannot do substring matching, ca_grants_by_agency resolves the caller's fragment against the 73 distinct AgencyDept values in one projected scan, then filters server-side on the exact names it matched.
Quick Start
Add to your MCP client (Claude Desktop, Cursor, Windsurf, etc.):
{
"mcpServers": {
"ca-grants": {
"url": "https://gateway.pipeworx.io/ca-grants/mcp"
}
}
}
Or connect to the full Pipeworx gateway for access to all 1375+ data sources:
{
"mcpServers": {
"pipeworx": {
"url": "https://gateway.pipeworx.io/mcp"
}
}
}
Using with ask_pipeworx
Instead of calling tools directly, you can ask questions in plain English:
ask_pipeworx({ question: "your question about Ca Grants data" })
The gateway picks the right tool and fills the arguments automatically.
More
License
MIT
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。