MCP search server for AI agents.
Give your AI agent searchable cloud storage in one config block. Sprigr connects via the Model Context Protocol so agents can store JSON objects, build indexes, and run full-text or hybrid semantic search queries. No custom code, no vector database, no infrastructure to manage.
Cloud-hosted: data available from any machine or session. Full-text search with optional hybrid semantic: keyword precision plus AI-powered meaning matching. Zero code: add one MCP config block. Unlimited queries, flat monthly pricing. Multi-tenant with isolated indexes per project and API key scoping.
You focus on the work. Sprigr runs the paperwork.
- MCP native
- Hybrid semantic search
- Scoped API keys
- Flat pricing
Built for agents, not afterthoughts
Every feature is designed for how AI agents actually work with data.
-
Zero-config cloud storage
No provisioning, no database setup, no migrations. Sign up, get an API key, paste the MCP config. Your agent can start storing and searching data within seconds.
-
Full-text & hybrid semantic search
Deterministic keyword search with typo tolerance, prefix matching, and custom ranking. Enable
semantic_searchon any index to add AI-powered vector search. Results are merged via Reciprocal Rank Fusion for the best of both worlds. -
Multi-tenant isolation & ACL
Each API key scopes access to its own data, with optional index-level ACLs (exact names or prefix wildcards) to restrict keys to specific indexes. For per-end-user access, mint scoped keys that carry an engine-enforced locked filter, so one key can only ever see its own rows. No cross-contamination, no shared state.
-
Flat, predictable pricing
No per-query fees, no per-token surcharges, no surprise bills when your agent runs a hundred searches in one session. One monthly price based on record count.
-
REST API + MCP
The same data is accessible via both MCP tool calls and a standard REST API. Use MCP for agent workflows, REST for dashboards, scripts, or non-MCP integrations.
-
Edge-deployed Rust backend
Sprigr runs across 300+ edge locations globally. The backend is compiled from Rust for native performance wherever your agent runs.
How MCP search works with Sprigr
Three steps from zero to searchable data inside your AI agent.
- 01
Connect
Add Sprigr to your AI client's MCP config. One JSON block with your API key and endpoint. No SDK to install, no server to run.
- 02
Store
Your agent pushes JSON objects to Sprigr via MCP tool calls. Define searchable attributes and filterable fields. Data is indexed automatically on write.
- 03
Search
Run full-text queries with typo tolerance, field filters, and pagination. Results return in milliseconds from the nearest edge node.
Add "sprigr": { "command": "npx", "args": ["-y", "@sprigr/mcp-server"], "env": { "SPRIGR_API_KEY": "your-api-key" } } to your MCP config. Your agent can create indexes, push records, and search through natural conversation. Knowledge base agents, conversation memory, and multi-agent coordination all built from the same primitive.
Questions
What is MCP and why does it matter for search?
MCP (Model Context Protocol) is an open standard created by Anthropic that lets AI assistants connect to external tools and data sources through a uniform interface. Instead of writing a custom API integration for every service, an AI agent can use MCP tool calls to interact with any MCP-compatible server. Sprigr exposes search, indexing, and storage as MCP tools, so your agent gets persistent, searchable memory without any custom code.
Which AI clients support the Sprigr MCP server?
Any MCP-compatible client works. Claude Desktop, Claude Code, the MCP SDK in Python and TypeScript, and any custom agent built on the MCP standard. The configuration format is the same across clients: a single JSON block pointing at the Sprigr server with your API key.
Do I need to manage embeddings or vector databases?
No. Sprigr handles embeddings internally when you enable semantic search on an index. You push raw JSON records, Sprigr generates and stores the vectors, and search queries use hybrid keyword + semantic ranking automatically. No embedding model to run, no vector database to provision.
How is data isolated between different agents or projects?
Each API key can be scoped to specific indexes using index-level ACLs. Create one API key per agent or project with access only to the relevant indexes. For multi-user apps, mint a secured key per end user: it carries a locked filter the engine ANDs into every query, on both keyword and semantic paths, so the holder can never widen access. Data is isolated per tenant and queries from one key can never access data belonging to another.
Is there a free tier for experimenting?
Yes. Sprigr has a permanent free tier that includes 1,000 objects, 1 index, and unlimited search queries. You can experiment with MCP integration, prototype agent workflows, and even ship small production projects without paying anything. Paid plans start at $49/month for larger indexes.
Give your agent searchable memory.
Paid plans from $49 a month.