Wraps Grants.gov and SAM.gov into eight typed tools that let agents search open federal grants by keyword and amount, find contracts filtered by NAICS code or set-aside type, pull full opportunity details, and scan upcoming deadlines. The match_opportunities tool takes an org profile and returns ranked results across both sources. Responses are normalized JSON with 24-hour SQLite caching so repeated queries don't hammer upstream APIs. Ships with a 50-call-per-day free tier and a $19/month unlimited pro option. Solid choice if you're building a procurement copilot, grant finder, or B2G sales tool that needs structured access to the full federal funding landscape without writing XML parsers or dealing with agency rate limits directly.
Public tool metadata for what this MCP can expose to an agent.
complete_profileGrantIQ complete_profile: fills in the user's matching profile field-by-field via chat. Partial-update semantics — null/missing fields leave existing values untouched. Server filters beneficiary/mission tags to canonical taxonomies. After this returns saved=true, call find_opp...1 paramsGrantIQ complete_profile: fills in the user's matching profile field-by-field via chat. Partial-update semantics — null/missing fields leave existing values untouched. Server filters beneficiary/mission tags to canonical taxonomies. After this returns saved=true, call find_opp...
inputobjectsearchGrantIQ search: find UK & EU grants matching a free-text query. Returns a ranked list of grant references (id, title, funder, deadline, amounts) the caller can pass to the `fetch` tool for full detail. Results are limited to currently-open, AI-enriched grants — closed, suspend...2 paramsGrantIQ search: find UK & EU grants matching a free-text query. Returns a ranked list of grant references (id, title, funder, deadline, amounts) the caller can pass to the `fetch` tool for full detail. Results are limited to currently-open, AI-enriched grants — closed, suspend...
limitintegerquerystringexplain_opportunityGrantIQ explain_opportunity: plain-English unpack of a single grant. Returns a summary, what the grant funds, target applicant profile, hidden constraints (e.g. state aid, consortium requirements), application mechanics (stages, duration, key dates), structured eligibility ess...1 paramsGrantIQ explain_opportunity: plain-English unpack of a single grant. Returns a summary, what the grant funds, target applicant profile, hidden constraints (e.g. state aid, consortium requirements), application mechanics (stages, duration, key dates), structured eligibility ess...
grant_idstringshould_applyGrantIQ should_apply: opinionated apply / don't-apply decision for a specific grant. Returns a verdict (yes / borderline / no), confidence (0-1), specific blockers, an effort estimate with confidence band, competition-risk label (or 'unknown' when funder data is absent), and t...1 paramsGrantIQ should_apply: opinionated apply / don't-apply decision for a specific grant. Returns a verdict (yes / borderline / no), confidence (0-1), specific blockers, an effort estimate with confidence band, competition-risk label (or 'unknown' when funder data is absent), and t...
grant_idstringfetchGrantIQ fetch: return the full record for a single grant by id. The id must come from a prior `search` call. Returns 'not found' when the grant is unknown OR not currently open/visible — there is no leak between those cases.1 paramsGrantIQ fetch: return the full record for a single grant by id. The id must come from a prior `search` call. Returns 'not found' when the grant is unknown OR not currently open/visible — there is no leak between those cases.
idstringfind_opportunitiesGrantIQ find_opportunities: runs the user's profile through the AI matching engine to find the top grants. Uses the canonical StagedMatchingService. Requires a complete profile (Location, OrgType, etc.).1 paramsGrantIQ find_opportunities: runs the user's profile through the AI matching engine to find the top grants. Uses the canonical StagedMatchingService. Requires a complete profile (Location, OrgType, etc.).
limitintegerA paid MCP server that gives AI agents instant, structured access to every open US federal grant and contract opportunity. Backed by Grants.gov + SAM.gov, with 24-hour SQLite caching, per-key daily limits, and a Stripe-friendly upgrade hook.
https://mcp-grantiq.up.railway.app/mcp
Title: GrantIQ MCP — US Federal Grants & Contracts for AI Agents
Subtitle: Search, match, and monitor every open SAM.gov and Grants.gov opportunity from any LLM, in one tool call.
GrantIQ MCP turns the messy world of US government funding into a clean, agent-ready API. Eight typed tools cover everything from broad keyword search and NAICS-coded contract discovery to deadline radar and historical award lookups. Responses are normalized across both data sources, returned as plain JSON, and cached for 24 hours so repeat queries don't burn rate limit on the upstream agencies.
Built for procurement copilots, grant-finder agents, and B2G sales tools, GrantIQ MCP ships with a free tier (50 calls/day) that any developer can drop into Claude or Cursor in 15 seconds. Upgrade to Pro for $19/mo for unlimited calls, or pay $0.02 per call for bursty workloads — billing flips a key from free to pro via a single webhook, with no app rewrites required.
| Tier | Price | Limit |
|---|---|---|
| Free | $0 | 50 calls / day / key |
| Pro | $19 / month | Unlimited |
| Pay-as-you-go | $0.02 / call | Unlimited |
claude mcp add grantiq-mcp --url https://mcp-grantiq.up.railway.app/mcp
Then add the API key Claude should use:
claude mcp env grantiq-mcp X-API-Key=grantiq-dev-key-001
The dev key grantiq-dev-key-001 is seeded automatically and capped at 50 calls/day. Mint your own with the admin endpoint below.
| Tool | Args | Returns |
|---|---|---|
search_grants | keyword, agency, amount_min, amount_max, status, limit | List of matching grants |
get_grant_details | opportunity_id | Full grant object |
search_contracts | keyword, naics_code, agency, set_aside, limit | List of SAM contracts |
get_contract_details | notice_id | Full contract object |
match_opportunities | org_description, org_type, focus_areas | Top 10 matched opps (grants + contracts) |
get_agencies | type (grants / contracts / both) | Agencies with open opps, ranked by volume |
get_deadlines | days_ahead | Opportunities closing within N days |
search_awards | recipient_name, agency, year | Past SAM award history |
Every tool returns either:
{ "success": true, "cached": false, "data": { ... } }
or, on failure:
{ "success": false, "error": "..." }
When the daily limit is hit, the HTTP layer returns 429 with body:
{ "error": "Limit reached. Upgrade at mcpize.com/grantiq-mcp" }
The MCP endpoint speaks the JSON-RPC streamable-HTTP protocol and is best used through an MCP client. For convenience, every tool also has a REST mirror at /tools/<tool>.
# Search grants
curl -s https://mcp-grantiq.up.railway.app/tools/search_grants \
-H 'Content-Type: application/json' \
-H 'X-API-Key: grantiq-dev-key-001' \
-d '{"keyword":"renewable energy","limit":5}'
# Get grant details
curl -s https://mcp-grantiq.up.railway.app/tools/get_grant_details \
-H 'Content-Type: application/json' \
-H 'X-API-Key: grantiq-dev-key-001' \
-d '{"opportunity_id":"345678"}'
# Search SAM.gov contracts (cyber, small-business set-aside)
curl -s https://mcp-grantiq.up.railway.app/tools/search_contracts \
-H 'Content-Type: application/json' \
-H 'X-API-Key: grantiq-dev-key-001' \
-d '{"keyword":"cybersecurity","set_aside":"SBA","limit":5}'
# Get contract details
curl -s https://mcp-grantiq.up.railway.app/tools/get_contract_details \
-H 'Content-Type: application/json' \
-H 'X-API-Key: grantiq-dev-key-001' \
-d '{"notice_id":"abc123"}'
# Match opportunities to an org profile
curl -s https://mcp-grantiq.up.railway.app/tools/match_opportunities \
-H 'Content-Type: application/json' \
-H 'X-API-Key: grantiq-dev-key-001' \
-d '{"org_description":"Battery storage R&D startup","org_type":"for-profit","focus_areas":["energy","battery","storage"]}'
# Top agencies right now
curl -s https://mcp-grantiq.up.railway.app/tools/get_agencies \
-H 'Content-Type: application/json' \
-H 'X-API-Key: grantiq-dev-key-001' \
-d '{"type":"both"}'
# Deadlines in the next 14 days
curl -s https://mcp-grantiq.up.railway.app/tools/get_deadlines \
-H 'Content-Type: application/json' \
-H 'X-API-Key: grantiq-dev-key-001' \
-d '{"days_ahead":14}'
# Past award history
curl -s https://mcp-grantiq.up.railway.app/tools/search_awards \
-H 'Content-Type: application/json' \
-H 'X-API-Key: grantiq-dev-key-001' \
-d '{"recipient_name":"Acme Corp","year":2025}'
Mint a fresh key (free or pro):
curl -s https://mcp-grantiq.up.railway.app/admin/keys \
-H 'X-Admin-Token: $GRANTIQ_ADMIN_TOKEN' \
-H 'Content-Type: application/json' \
-d '{"tier":"free","daily_limit":50}'
Promote a free key to pro (called from the Stripe checkout success webhook):
curl -s https://mcp-grantiq.up.railway.app/billing/upgrade \
-H 'X-Admin-Token: $GRANTIQ_ADMIN_TOKEN' \
-H 'Content-Type: application/json' \
-d '{"key":"grantiq_xxxxx"}'
python -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt
cp .env.example .env
python server.py # http://localhost:8000
pytest -v # 22 tests, hermetic with respx mocks
The repo includes nixpacks.toml. After railway init, just railway up. The start command is python server.py (not uvicorn ...) because Railway exec's [start].cmd without a shell — server.py reads $PORT via os.getenv and starts uvicorn programmatically.
Set these on the Railway service:
SAM_API_KEY — your SAM.gov keyGRANTIQ_ADMIN_TOKEN — long random stringGRANTIQ_DB_PATH — /data/grantiq.db if you mount a Volume (recommended so cache + keys survive deploys)A ready-to-run deploy.sh is in this directory.
ptype=a (Award Notice). For deeper award analytics (CFDA-level totals, recipient rollups), USAspending.gov is more appropriate and would be a v2 add./admin/keys to mint a key that bypasses cache (not yet exposed — see TODO)./billing/upgrade). The MCP server doesn't host the checkout page itself — that lives on the marketing site.GRANTIQ_API_KEY*secretGrantIQ API key (free dev key `grantiq-dev-key-001` or mint via /admin/keys).
io.github.ericm1018/skillfm-llm-cost-optimizer-openai-anthropic-usage
io.github.mikerawsonnz/llm-orchestration-agent
io.github.mikerawsonnz/authenticated-llm-agent
labforgedev/copilot-memory-mcp
csoai-org/agent-prompt-injection-firewall-mcp
io.github.mikerawsonnz/authenticated-multi-llm-agent