SparkFeed

API & MCP

Your feeds,
ready for tools.

Connect a Sparkfeed workspace through the REST API or MCP. List feeds, search articles, and request article text with a workspace-scoped key.

Editorial collage of source pages passing through a blue opening into a collected book
One workspace REST API /api/v1 MCP /api/mcp

One workspace. Two ways to connect.

Use direct HTTP requests or give a compatible AI client MCP tools. Both connect to the same Sparkfeed workspace and tool definitions.

Give an agent the sources you chose.

MCP connects a compatible AI client to feeds already in your workspace. It can list sources, search article titles and summaries, then request one article's text.

Read the MCP setup guide

Use REST for direct requests.

Send standard HTTP requests and receive JSON. The REST API and MCP tools use one registry, so their names, inputs, and access rules stay aligned.

Browse the API reference

Set access with workspace keys.

Create a key for your integration and choose its scopes. Read tools and write actions are available only when the key carries the matching access.

See API key scopes

Give each request a clear job.

Start with reading. Add write access only when your workflow needs it.

Find an article

Search by text, folder, feed, date, or reading state.

Request article text

Get one article as markdown by default. The response identifies the content source.

Scope each key

Keep workspace, article, and write permissions tied to the key.

Check the API spec

Fetch the public OpenAPI document before you build an integration.

A direct connection to your workspace

Bring your sources
into the workflow.

Start with the feeds you already follow. Connect them through the API or MCP when your tools need them.

01What can I read through the API and MCP?
List folders and feeds, search articles, and request one article's text. The response says whether Sparkfeed read the page, returned its RSS description, or could not read it.
02How does the MCP connection work?
MCP clients connect to /api/mcp over Streamable HTTP and send a Sparkfeed API key as a bearer token. Available tools follow the key's scopes.
03Can I use the REST API without an AI agent?
Yes. Send JSON requests to /api/v1 with a Sparkfeed API key. The public OpenAPI document lists the current endpoints and parameters. Read the API reference.
04Can any API key change my workspace?
Only a key with the required write scope can use those tools. The public demo key is read-only.
05Does Sparkfeed fact-check an AI model's answer?
The API and MCP provide workspace data. The connected model handles its own reasoning, so Sparkfeed does not verify the model's conclusions.