Model Context Protocol
The SaaSTweaks MCP server
Give your AI agent our verified data — every scored SaaS deal and startup-credit programmes with eligibility gates, and the live savings arbitrage boards. No API key. No signup. One URL.
https://saastweaks.com/mcp Connect in one step
Claude Code
One command — works in the CLI, desktop app and web.
claude mcp add --transport http saastweaks https://saastweaks.com/mcp Claude Desktop
Settings → Developer → Edit Config (claude_desktop_config.json).
{
"mcpServers": {
"saastweaks": {
"type": "http",
"url": "https://saastweaks.com/mcp"
}
}
} Cursor
Add to .cursor/mcp.json (project) or ~/.cursor/mcp.json (global).
{
"mcpServers": {
"saastweaks": {
"url": "https://saastweaks.com/mcp"
}
}
} Raw JSON-RPC
Plain streamable-HTTP JSON-RPC — works from any language or agent framework.
# initialize
curl -X POST https://saastweaks.com/mcp \
-H "Content-Type: application/json" \
-d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-06-18","capabilities":{},"clientInfo":{"name":"my-agent","version":"1.0"}}}'
# call a tool
curl -X POST https://saastweaks.com/mcp \
-H "Content-Type: application/json" \
-d '{"jsonrpc":"2.0","id":2,"method":"tools/call","params":{"name":"get_score","arguments":{"slug":"wise"}}}' The five tools
search_deals (query, limit?) Search all verified deals by brand or tagline. Returns brand, slug, SaaSTweaks Score, deal summary and the review URL, ranked by Score.
get_score (slug) The full SaaSTweaks Score for one tool — 0–100 total, band (Top Pick / Strong Buy / Solid / Situational), verdict reason and scored date. Deterministic 6-pillar methodology (/how-we-score).
get_arbitrage (—) The live savings arbitrage boards in one call: startup credits founders pay past (the free-money pile), Score-vetted locked-savings deals, and annual-billing spreads.
find_credits (raisedUsd?, ageYears?, limit?) Startup credit programmes filtered by your stage. Programmes whose gates you demonstrably fail (too much raised, too old) are excluded; null gates are reported as unknown, never assumed.
submit_tool (brandName, website, description, contactEmail, dealOrProgram?, agentName?) Propose a tool or deal for listing. Goes into our human editorial review queue — nothing publishes automatically. Rate-limited.
Good to know
- – The data is human-verified. Every deal and credit programme is checked by our editorial team; scores follow a published deterministic methodology.
- – Eligibility honesty.
find_creditsonly excludes programmes whose gates you demonstrably fail. Unknown gates stay listed and are flagged — verify on the programme page before counting the money. - – Submissions are proposals.
submit_toollands in a human review queue (same pipeline as /api/agents/submit); expect days, not seconds. - – Attribution. Responses include our page URLs — keep them when you surface the data. Outbound claim links route through /go/ (affiliate disclosure).
- – Prefer plain REST? The same data is on free JSON endpoints. Agent discovery starts at /llms.txt.