spacefrontiers_search_documentsLABSCO SUMMARY
Formal literature and social sources kept behind separate tools, with full text bounded rather than cut off arbitrarily.
Splitting papers from Reddit is the useful decision: the agent states which kind of evidence it wants instead of hoping a single ranker guesses right. Retrieval is staged as well — a search returns snippets with the estimated full-text size attached, and only then do you fetch the document under an explicit character budget or search inside it for the handful of passages that matter. Every call has a published price, which is also why the citation-backlink option is off by default.
What it is
A hosted retrieval layer for agents over two different bodies of material: the formal literature — peer-reviewed papers, books, patents, standards and Wikipedia — and social sources such as Reddit, Telegram channels and YouTube transcripts. Every result carries a canonical URI, so a claim can be cited back to its source.
What you get
- spacefrontiers_search_documents covers the formal literature for citations and prior art, while spacefrontiers_search_social covers Reddit, Telegram and YouTube transcripts for news and community discussion
- Every hit returns a canonical source_uri, a snippet of up to 900 characters, an abstract preview of up to 800, a relevance score, authors, date, type, and the estimated size of the full text
- spacefrontiers_fetch_document returns bounded full text for one URI — 40K characters by default and up to 100K — with up to 50 references; citation backlinks are opt-in because they trigger a separately billed search
- spacefrontiers_search_in_document returns up to five matching passages inside a single document, which is the route to take for anything past roughly 20K tokens
- Search returns 10 hybrid-ranked results by default and is capped at 30
- All four tools are read-only and idempotent, declare readOnlyHint, idempotentHint, openWorldHint and destructiveHint false, and are prefixed spacefrontiers_ so they do not collide in a multi-server setup
- A deep_research_agent prompt and a spacefrontiers://document/{uri_b64} resource template ship alongside the tools
Requirements
A hosted Streamable HTTP server at https://mcp.spacefrontiers.org/. Authentication is OAuth 2.1 with PKCE and dynamic client registration — a browser opens on first use, with no key to paste — or a long-lived API key sent as a Bearer header. Self-hosting over stdio is possible from the repository, with SPACE_FRONTIERS_API_KEY as the upstream credential. Calls are billed per use: a search is $0.01 plus $0.001 per returned result, so the 10-result default costs $0.02; a full document fetch is $0.05; an in-document passage search is $0.015; and asking for citation backlinks adds another billed search.
Setup effort
One command — claude mcp add --transport http --scope user spacefrontiers https://mcp.spacefrontiers.org
USE CASES
01
Checking prior art before you build something
Search the papers, patents and standards, then pull the full text of the two or three that actually matter.
02
Reading only the relevant part of a long paper
Rather than fetching 100K characters, ask for the five passages inside the document that match the question.
03
Seeing what practitioners are actually saying
The social search covers Reddit threads, Telegram channels and YouTube transcripts, where the formal literature has nothing yet.
04
Producing citations that resolve
Every hit carries a canonical source_uri, and the fetch returns the document behind it with up to 50 references.
05
Keeping a research loop's cost predictable
Results are priced per hit and capped at 30, so a broad search and a targeted fetch cost what the price list says.
4 tools
Listed in project docs. Your client may show a different list.
spacefrontiers_search_socialspacefrontiers_fetch_documentspacefrontiers_search_in_documentBEFORE YOU INSTALL
What to know first
Tools you get
4 tools named in the project's documentation.
Requires
A client that can reach a remote MCP server over HTTP
INSTALLATION
Connect your client
Server setup
claude mcp add
Claude Code
TERMINAL
claude mcp add --transport http --scope user spacefrontiers https://mcp.spacefrontiers.orgHosted endpoint
ENDPOINT URL
https://mcp.spacefrontiers.orgAdd this hosted endpoint to your MCP client.
MCP Clients
Cursor
CURSOR · JSON
{
"mcpServers": {
"spacefrontiers": {
"type": "http",
"url": "https://mcp.spacefrontiers.org",
"headers": {
"Authorization": "Bearer <YOUR_API_KEY>"
}
}
}
}VS Code
VS CODE · JSON
{
"servers": {
"spacefrontiers": {
"type": "http",
"url": "https://mcp.spacefrontiers.org",
"headers": {
"Authorization": "Bearer <YOUR_API_KEY>"
}
}
}
}Cline
CLINE · JSON
{
"mcpServers": {
"spacefrontiers": {
"type": "streamableHttp",
"url": "https://mcp.spacefrontiers.org",
"headers": {
"Authorization": "Bearer <YOUR_API_KEY>"
}
}
}
}Windsurf
WINDSURF · JSON
{
"mcpServers": {
"spacefrontiers": {
"type": "http",
"serverUrl": "https://mcp.spacefrontiers.org",
"headers": {
"Authorization": "Bearer <YOUR_API_KEY>"
}
}
}
}Any MCP client — self-hosted
ANY MCP CLIENT — SELF-HOSTED · TERMINAL
SPACE_FRONTIERS_API_KEY=<YOUR_SPACE_FRONTIERS_API_KEY> uv run fastmcp run mcp_server.py
