buzzabout docsMCP overview
How buzzabout exposes itself as an MCP server — the assistant flow, the read tools, and how to authenticate.
The buzzabout MCP server lets any Model Context Protocol client — Claude Desktop, Claude.ai, Claude Code, Codex, Cursor, ChatGPT, your own SDK agent — drive buzzabout from inside the conversation. You ask a research question; the assistant collects and analyses the posts and returns the answer in the chat, often as an interactive widget.
Transport
https://mcp.buzzabout.ai/mcp/Streamable HTTP — one URL handles tool listing, tool calls, and OAuth discovery (no separate SSE endpoint).
Trailing slash is required
Use https://mcp.buzzabout.ai/mcp/ (with trailing slash). The
unslashed /mcp returns a 307 redirect that strips the request body
in many MCP clients, which surfaces as silent connection failures or
empty tool lists.
See Use in your agent for per-host wiring.
Authentication
Two paths on the same URL, chosen by client:
- OAuth — standard assistants (Claude Desktop, Claude.ai, ChatGPT) handle it; just sign in.
x-api-key— CLI / IDE agents (Claude Code, Codex, Cursor) and custom agents paste a key.
Both resolve to the same buzzabout account. Full details on Authentication.
What's exposed
15 tools, in two kinds:
- The assistant flow —
ask→get_message→render.askhands a prompt to the buzzabout assistant, which writes the query, previews it, collects posts, profiles audiences, detects patterns, and analyses. It returns immediately; you pollget_messagefor the answer, andrenderdraws any rich block as an interactive widget. - Read-only lookups — list / fetch datasets, runs, mentions, audience profiles, and tracking agents the account already has.
There are no create / update / delete tools — collecting,
profiling, and pattern detection all happen through ask. For
programmatic CRUD, use the REST API.
Tools reference
Every tool — the assistant flow plus the read-only lookups.
Use in your agent
Per-host wiring + a first prompt to test the loop.
Authentication
OAuth for standard assistants, x-api-key for Claude Code / Codex / Cursor.
The async flow
Research can run for minutes — past a host's tool-call timeout — so the assistant flow is asynchronous:
buzzabout__ask(prompt)returns immediately with{ chat_id, message_id, status: "working" }.- The host polls
buzzabout__get_message(chat_id, message_id)— which long-polls (one call holds for ~45s, returning the moment the turn settles) — untilstop_reasonis non-null. - The answer comes back as
blocks: arender: trueblock is shown viabuzzabout__renderas an interactive MCP App widget; arender: falseblock carries plain markdown the host relays. Hosts without the MCP Apps UI extension get markdown throughout.
Credit-exhausted accounts surface a 402 with a structured "need N more
credits" body so the assistant can ask the user to top up before
retrying.
When to use MCP vs REST
| Use MCP | Use REST |
|---|---|
| Interactive — a person talking to an LLM client. | Batch — scheduled job, cron-driven sync. |
| You want the assistant to drive the research. | You're writing the orchestration yourself. |
| Results inline in the conversation. | You need primitive CRUD / precise control. |
Both surfaces are backed by the same primitives, so hybrid setups work naturally (e.g. run a heavy collection via REST, then ask an MCP-capable assistant to summarise it).
See also
- Agentation — third-party MCP devtool that pairs well with buzzabout for local agent development and inspection.
Next steps
- Tools reference — the 15-tool surface.
- Use in your agent — per-host wiring + first prompt.
- Authentication — pick OAuth or
x-api-key.
Create a watchlist
Keep a fixed set of competitors, influencers, subreddits, or URLs under continuous watch by building a URL-based dataset and attaching a listening agent focused on what matters.
Use in your agent
Wire buzzabout's MCP server into any MCP-capable client — Claude, Claude Code, Codex, Cursor, ChatGPT, or your own SDK-built agent.