Grant data for AI agents and apps.
Semantic grant search, funder profiles, and award history over a corpus of 5,000+ verified-open opportunities — not an inflated catalog of duplicates and expired listings. A REST API plus a Model Context Protocol server for Claude, Cursor, and any MCP client.
Five lines, then your AI can find grants
Add the MCP server to your client config and restart. Your assistant can now search the open-grant corpus, pull funder profiles, and check eligibility — in the flow of a normal conversation. The free tier needs no key.
- Works with Claude Desktop, Cursor, and any MCP client
- Streamable HTTP — no local install
- Read-only and idempotent: safe for autonomous agents
{
"mcpServers": {
"grantsonar": {
"url": "https://grantsonar.com/api/mcp"
}
}
}"headers": { "Authorization": "Bearer gsk_…" } with a key from Settings → API & MCP.Add it to your client
The free tier needs no key. For higher limits, add an Authorization: Bearer header with a key from Settings → API & MCP.
claude mcp add --transport http grantsonar https://grantsonar.com/api/mcp
https://grantsonar.com/api/mcp
{
"mcpServers": {
"grantsonar": { "url": "https://grantsonar.com/api/mcp" }
}
}{
"mcpServers": {
"grantsonar": { "serverUrl": "https://grantsonar.com/api/mcp" }
}
}[mcp_servers.grantsonar] url = "https://grantsonar.com/api/mcp"
The tools
Read-only tools your agent can call directly. More of the graph is rolling out.
search_grantsLiveSemantic + keyword search over the catchable grant corpus. Filter by state, source, funder, eligibility. Returns deadlines, award amounts, and a GrantSonar link.
get_grantLiveFull detail for one opportunity by id — synopsis, eligibility, deadlines, award floor/ceiling, application URL.
search_funders / get_funderSoon230K+ foundation profiles — giving focus, recent grants, financials, and leadership. The funder side of the graph.
get_past_winnersSoonFederal award history — who actually won, how much, and where. Competitive intelligence on every program.
match_grants_to_orgSoonScore and rank opportunities against an organization profile (mission, focus, geography, eligibility). Our matching engine, not just keyword search — most grant APIs do not have this.
Why build on GrantSonar
Honest, verified-open data
We return 5,000+ grants that are actually open right now — not an inflated catalog of duplicates and expired listings. Eligibility-aware and freshness-checked against the source.
Matching, not just search
match_grants_to_org scores opportunities against a real org profile. Hand your agent a mission and get ranked, eligible results — the thing keyword APIs can’t do.
Read-only and fast
Every tool is read-only and idempotent. No write access, no surprises — safe to hand to an autonomous agent. Semantic search returns in tens of milliseconds.
Simple, metered pricing
Start free, no signup. Pay only when you scale — and the per-call rate drops as you grow.
For evaluation and AI-assistant use. Anti-abuse daily cap, shared fair-use.
Add to ClaudeYour own key + burst capacity for side projects and batch pulls.
Start checkoutProduction apps. 60% more volume than the next platform at a third less.
Start checkoutScaled usage with the best per-call rate. Priority support.
Start checkoutBulk exports, custom datasets, SLA. Still far below legacy data vendors.
Contact usThe free tier is live now, no signup. Paid plans use hard monthly caps today; the $/1K figures are effective-rate comparisons, not metered overage charges. Every tier undercuts other grant APIs on price and volume, and runs far below opaque enterprise data vendors. Metered overages and AI-heavy operations, if added, are future options.
Developer FAQ
No. The free tier is unauthenticated — drop the endpoint into your MCP client and start searching. Create a free account and generate a key only when you want higher limits.
Federal, state, and foundation grant opportunities that are currently open, plus 230K+ funder profiles and federal award history. We verify open status against the source rather than padding the count with closed listings.
Free is a per-day cap for evaluation. Paid plans are monthly query quotas with burst capacity. AI-heavy operations (if added later) are metered separately so a runaway agent can’t blow through your plan.
No. Every tool is read-only and idempotent. It is safe to expose to an autonomous agent — the worst it can do is search.
Self-serve, transparent pricing, and a verified-open corpus — at a fraction of the cost of opaque enterprise data vendors, and undercutting other grant APIs on both price and volume.
Point your agent at the real open-grant graph.
Free to start, no signup. Five lines of config and your assistant can find grants that are actually open.
Get started