Conference Partner API & MCP server

Programmatic, read-only access to the whole Conference Partner dataset: 5,799 academic conferences and 1,213 journals, with Calls for Papers and submission deadlines, conference dates and locations, CCF / ICORE / QUALIS rankings side by side, historical acceptance rates, edition history, journal impact factors, ISSNs and special issues. Two interfaces over the same data: a REST API, and a native MCP server for AI agents.

Try it right now

These three commands work as-is, with no credentials and no account:

# AI-field conferences with a submission deadline in the next three months
# (no keyword needed — every search parameter is optional and combinable)
curl "https://www.myhuiban.com/api/conferences?field=ai&submission_date_start=2026-08-01&submission_date_end=2026-11-01"

# Conferences whose call for papers is currently open
curl "https://www.myhuiban.com/api/conferences/rankings/cfp?per_page=5"

# Site-wide statistics (totals, deadlines by month, ranking tiers, research fields, countries)
curl "https://www.myhuiban.com/api/statistics"

Every response carries X-Quota-Limit and X-Quota-Remaining so you can see what is left for the day. CORS is open, so these also work from browser JavaScript.

REST endpoints

Paginate any list with page and per_page (max 100). Full machine-readable contract: OpenAPI 3 schema.

MCP server (for AI agents)

A native Model Context Protocol server speaks Streamable HTTP at POST https://www.myhuiban.com/mcp, and is listed in the official MCP registry as com.myhuiban.www/conference-partner. Add it to any MCP client, for example:

claude mcp add --transport http huiban https://www.myhuiban.com/mcp

Eight tools: search_conferences, search_journals, list_conferences, list_journals, get_conference, get_journal, get_researcher, get_statistics.The search tools take the same optional filters as the REST search endpoints (research field across 21 areas, ranks, and submission / conference date ranges).

These work with no credentials: search_conferences, search_journals, list_conferences, list_journals, get_statistics. The three detail tools (get_conference, get_journal, get_researcher) need a credential — see below.

The server also ships MCP prompts — ready-made tasks your client offers as one-click starting points, so you do not have to work out what to ask a list of tool names: upcoming_deadlines, where_to_submit, venue_snapshot, open_special_issues. Each expands into a request that names the tools to call, and each takes optional arguments such as a research field or how far ahead to look.

Every MCP tool has a REST twin (get_researcher ↔ GET /api/researchers/{id}, and so on): the two surfaces run the same query layer and return identical payloads, so you can switch transports without changing how you parse anything.

Getting a credential for detail endpoints

1. Register a free account at myhuiban.com/registration and verify your email.
2. Create an API key at myhuiban.com/account/api-keys. It is shown once — we only store a hash of it.
3. Send it on every request: Authorization: Bearer <credential>

   curl -H "Authorization: Bearer hb_..." \
     https://www.myhuiban.com/api/conferences/1

This is also how you connect an MCP client. Claude, ChatGPT, Cursor and VS Code let you add a remote MCP server by URL (https://www.myhuiban.com/mcp) plus one fixed request header — an API key fits that form exactly, with no token-refresh round trip.

API keys never expire and can be revoked individually from the same page; each one is metered separately, so you can give every client its own. If you would rather exchange credentials at runtime, POST /api/login still returns a JWT valid for about six months — it works everywhere a key does, but cannot be revoked on its own.

Every credential on an account draws on the same quota (200 requests per day), shared across the REST API and the MCP server.

Three things are not self-serve and are available on request: a higher quota, each venue's own website URL (responses otherwise link only to detail_page), and a bulk export of the dataset. Email [email protected] with what you are building and we will tell you what is possible — there is no published price list, because so far every request has been different.

Keeping a copy in sync

Do a full pass once (no updated_since), then poll with updated_since set to the newest updated_at you saw. Only venues changed on or after that date come back.

curl 'https://www.myhuiban.com/api/conferences?updated_since=2026-08-01&per_page=100'

Two things to know: the resolution is one day, so re-fetching the newest date is normal and idempotent; and venues that have never been edited carry no change record, so they appear only in the full pass, never in the incremental feed.

Quotas and status codes

• 200 — success. Check X-Quota-Remaining for what is left today.
• 401 — this endpoint needs a credential (only the per-venue detail endpoints do), or the one you sent is invalid, revoked or expired. Send no Authorization at all to use the anonymous tier — an unusable credential is never silently downgraded to it.
• 402 — your account's daily quota is used up. Retry-After says how many seconds until it resets.
• 429 — the anonymous per-IP daily quota is used up; a free API key raises the limit a lot. Retry-After says how many seconds until it resets — wait that long instead of retrying sooner, and use updated_since so a scheduled job re-fetches only what changed.

Client examples and a downloadable snapshot

github.com/sundou82/conference-partner-mcp is the open companion repository: copy-paste MCP configuration for Claude, Cursor and VS Code, runnable Python / Node / curl clients that need no API key, and the OpenAPI document re-fetched from this site every day. The MCP server and the API themselves are a hosted service and are not open source — nothing in the repository has to be installed to use them.

It also carries a snapshot rebuilt daily, as JSON and CSV: upcoming deadlines, the CCF and CORE catalogues, CCF journals, open special-issue calls and journals by impact factor. That covers exactly what the endpoints above serve without credentials — per-venue detail stays behind the API. A file is stale the moment a deadline moves, so use it to browse or bulk-load and call the API when it has to be right.

Terms of use

The API is read-only and free to use, including in commercial products, as long as you attribute Conference Partner and link back to the venue's page on this site — every response includes a detail_page URL for exactly that purpose. Please keep one request per second or slower when crawling lists. For redistributing the dataset in bulk, use the published data snapshot rather than a crawl of your own, and get in touch first for anything wider — dataset licensing and model-training use in particular. See also our Methodology for where the rankings and data come from.

Other machine-readable entry points

• /api/openapi.json — OpenAPI 3 schema
• /llms.txt — site description for LLMs
• /llms-full.txt — site description for LLMs, plus live data samples
• /.well-known/api-catalog — RFC 9727 API catalog
• /auth.md — authentication reference
• A compact Markdown rendition of any conference or journal detail page — send an Accept: text/markdown request header (without text/html).
• RSS feed and iCalendar feed of upcoming deadlines
• github.com/sundou82/conference-partner-mcp — client examples, MCP configuration files and the daily data snapshot

Questions, bug reports or a use case you want us to support: [email protected].