savvly-mcp
<!--
CANONICAL PUBLIC README for the Savvly MCP server.
Source of truth lives here in savvly-public-api (private); the release workflow
(.github/workflows/publish-mcp-server.yml) syncs this file — plus server.json and
LICENSE — to the PUBLIC Savvly/savvly-mcp repo on every production release. Do not
edit it directly in savvly-mcp; edit here.
-->
# Savvly MCP server — `com.savvly/savvly`
Savvly offers longevity-linked, SEC-registered investment products — built on the Savvly 80+ fund — that invests pooled contributions in a low-cost S&P 500 ETF and makes milestone cash payouts at ages 80, 85, 90, and 95, turning a longer life into a financial reward rather than a risk.
Savvly MCP lets financial advisors, planning tools, and AI agents query fund data, model projections, and compare against alternative retirement products directly, in real time, in any conversation, and without any manual lookup.
Listed on the [MCP Registry](https://registry.modelcontextprotocol.io) as
`com.savvly/savvly` and published to **five package types plus a hosted remote** — use
whichever fits your client.
## Install
| Channel | Command / config |
| --- | --- |
| **Remote (hosted)** | `https://api.savvly.com/mcp` (streamable-http) |
| **npm** | `npx @savvly/mcp-server` |
| **PyPI** | `uvx savvly-mcp-server` |
| **NuGet** | `dnx Savvly.McpServer` (or `dotnet tool install -g Savvly.McpServer`) |
| **Docker / OCI** | `docker run -i --rm ghcr.io/savvly/savvly-mcp-server` |
| **MCPB** | download the `.mcpb` from [Releases](https://github.com/Savvly/savvly-mcp/releases) and open it in an MCPB-capable host |
### Claude Desktop / Cursor / Windsurf (stdio)
```json
{
"mcpServers": {
"savvly": { "command": "npx", "args": ["@savvly/mcp-server"] }
}
}
```
Swap `command`/`args` for `uvx savvly-mcp-server`, `dnx Savvly.McpServer`, or
`docker run -i --rm ghcr.io/savvly/savvly-mcp-server` as you prefer. Hosts that support
remote MCP can point straight at `https://api.savvly.com/mcp`.
## Tools
- `get_savvly_product_info` — full product overview
- `compare_savvly_vs_alternative` — Savvly vs annuities, target-date funds, etc.
- `check_savvly_eligibility` — age / residency / channel eligibility
- `get_savvly_faq` — FAQ by topic
- `search_savvly_content` — audience-tagged Q&A library
- `project_savvly_lumpsum` / `project_savvly_monthly` — payout projections (milestone ages)
- `project_retirement_with_savvly` — full retirement trajectory with vs. without Savvly
Projection tools emit interactive MCP Apps chart widgets on hosts that support them.
## Links
- Website: https://savvly.com
- Disclosures: https://www.savvly.com/disclosures
## Privacy Policy
Savvly's privacy policy: https://www.savvly.com/privacy-policy
This MCP server is public and unauthenticated — it exposes Savvly product information, comparisons, eligibility, and illustrative projections, and requires no account or credentials. It collects **no personal information** and **no end-user identifier** (no client IP is captured; no user is identified).
The hosted endpoint (`https://api.savvly.com/mcp`) records **anonymous usage analytics** via [AgentCat](https://agentcat.io) (formerly MCPcat), a third-party processor, to improve the tools — which tools are called and their non-identifying numeric scenario inputs (e.g. age, contribution amount). Email, phone, SSN, and payment-card patterns are redacted from any text before it is sent. The local `npx @savvly/mcp-server` (stdio) sends **no** analytics.
## License
Apache-2.0.
<!-- mcp-name: com.savvly/savvly -->
TDQS
Scored across 8 tools
The three projection tools (lumpsum, monthly, retirement_with_savvly) are separable by their input scenarios, but the informational cluster—get_savvly_product_info, get_savvly_faq, and search_savvly_content—overlaps heavily, with the FAQ tool even admitting it is a convenience view of search_savvly_content. Descriptions clarify boundaries reasonably well, but an agent could still pick the wrong info-retrieval tool.
All names use snake_case with a leading verb and the 'savvly' domain token (compare_savvly_vs_alternative, get_savvly_product_info, project_savvly_lumpsum, search_savvly_content). The pattern is highly predictable and readable throughout.
Eight tools is well-scoped for a product-information and retirement-projection server, with no redundant or filler tools. Each tool covers a distinct capability (info, comparison, eligibility, FAQ, search, and three projection variants).
The surface covers the core lifecycle of a retirement-product info server: educate (info/FAQ/search), compare, check eligibility, and project (lump sum, monthly, full trajectory). Minor gaps remain—no explicit expense/cost breakdown tool or enrollment/contact operation—but these are largely absorbed by the product info tool.