OpenGrants exposes an MCP server, so an AI assistant that supports the Model Context Protocol can use the grant database directly — searching, matching, checking eligibility and managing your pipeline inside a conversation.
What the agent can do
Read
- Search grants, government contracts, and the funder directory
- Pull the full record for a specific grant, contract or funder
- Run a full funding-landscape scan across all three at once
- Match open opportunities against your stored profile, with scores and rationales
- Check your eligibility for a specific grant
- Review your pipeline with deadline urgency
Write
- Save grants and contracts, and move them through pipeline statuses
- Create alerts
- Create and update your organization profile
Because some of those modify your account, it is worth knowing which is which — an agent saving things to your pipeline is doing real work on your data.
Set up the profile first
Store your organization profile once and every subsequent call personalizes automatically. Matching, eligibility checks and landscape scans all read from it, so the agent stops asking you to re-describe your mission on every request.
See Set up your organization profile.
Subscription requirements
Some capabilities are gated. Matching and eligibility checks require an active subscription, and eligibility specifically requires Pro. If your subscription lapses, those calls return a clear message pointing at pricing rather than degrading quietly.
Search and funder lookup are not gated the same way.
Good agent workflow
- Store the profile
- Landscape scan to see the whole picture
- Match for ranked fit
- Eligibility check on the top few
- Save the survivors to the pipeline
- Alert on the space so it keeps working
That sequence works well because each step narrows using real data rather than the agent guessing.
Choosing between MCP and the API
MCP is right for interactive agent sessions. For scheduled jobs and production integrations, use the REST API directly — token lifecycle and rate-limit behavior are much easier to reason about outside an agent transport.