API

Presspool's public read API

Presspool's funding weeks, niche maps, company records, and lead investors are available as JSON. The API is read-only. It needs no account, API key, or OAuth token.

Authentication

Send no credential. A GET request with no Authorization header is a complete request. auth.md says the same thing for agents: there is no registration, no token exchange, and no authorization server.

Endpoints

GET /api/v1 lists the endpoints. GET /api/v1/weeks returns every published week, and GET /api/v1/weeks/{slug} returns one. A slug looks like week-of-YYYY-MM-DD. GET /api/v1/companies returns every company, and ?q= filters by name, domain, or alias. GET /api/v1/companies/{domain} returns one company; the domain is the company's website, and the page slug also works. GET /api/v1/niches and GET /api/v1/niches/{slug} return the niche maps. GET /api/v1/investors and GET /api/v1/investors/{slug} return investors that led two or more rounds on the record. That list is not an investor's full portfolio. GET /api/v1/search?q= searches companies, niches, and investors in one response.

The same weeks, companies, and niches are at /weeks.json, /companies.json, and /niches.json. The machine-readable description is /openapi.json. The API catalog is at /.well-known/api-catalog.

Pagination

List reads keep the arrays clients already use: weeks, companies, niches, and investors. Each of those responses also has items, the same page, and nextCursor when another page exists. nextCursor is omitted when the response is the rest of the list. Omit limit and cursor to receive every record, which is what existing clients do. Pass limit (an integer from 1 to 100) and the opaque cursor from nextCursor to read a later page. A cursor without limit returns the records from that point through the end. /weeks.json, /companies.json, and /niches.json are the complete feeds and do not use this page shape.

Versioning and deprecation

The read API's current major version is v1, in the path /api/v1. A breaking change ships as a new major path, such as /api/v2. Adding a field is not a breaking change. /weeks.json, /companies.json, and /niches.json are unversioned copies of the same records. Removing or renaming a field on those feeds is breaking and follows this policy.

Nothing in v1 is deprecated, and nothing has a sunset date. When Presspool deprecates a surface, that surface's responses send Deprecation (RFC 9745) and Sunset (RFC 8594), and this section names the replacement and the sunset date before those headers are sent.

The MCP tools list_record_pages, get_record_page, and search_record_pages are the current names. list_funding_weeks, get_funding_week, list_companies, get_company, list_niches, get_niche, list_investors, and get_investor still call the same reads. Those aliases have no removal date.

Rate limits

The read API does not enforce an application rate limit. Responses do not send RateLimit, RateLimit-Policy, or Retry-After, because those headers would name a quota the server does not apply. A host in front of the site may still refuse a flood; that refusal is not a published Presspool quota.

Examples

A week list is curl https://presspool.ai/api/v1/weeks. One company is curl https://presspool.ai/api/v1/companies/example.ai, using that company's real domain. A filtered company list is curl "https://presspool.ai/api/v1/companies?q=example".

Errors

Errors are application/problem+json. Each body has type, title, status, detail, instance, code, and resolution. not_found means the record is not on Presspool. invalid_request means the slug, domain, or q parameter is not usable. method_not_allowed means the request was not a GET. The API does not create or change records.

MCP

A read-only MCP server at /mcp speaks Streamable HTTP. Its card is at /mcp/server-card. tools/list returns three tools, list_record_pages, get_record_page, and search_record_pages, each with a JSON Schema for its parameters and its result. The previous eight tool names still work on tools/call. In a browser that supports WebMCP, the same tools are registered on the page with document.modelContext.registerTool.

Markdown

Every editorial page returns Markdown for Accept: text/markdown, and at the page path plus .md. /llms.txt is the navigation index. This page's section index is /api/llms.txt.