Bilibili API for AI agents and LLMs: fewer tokens, called from Claude and Cursor (2026)

Why does an agent care how big an answer is?

Every token of a tool's answer goes into the agent's context and is paid for as input. Bilibili's raw JSON is made for its own app, with many fields nested deep. Across DataXAPI's 9 Bilibili endpoints, one recorded real response each, the raw JSON comes to 477k tokens and what an AI reads through MCP to 21k. With the context saved, an agent reads more pages and takes more steps in one conversation, and sooner.

Every platform uses the same names: the author in author, counts in metrics, the time in published_at as UTC, the next page in page.next_cursor. A field the platform does not give is listed in meta.unavailable, so the AI does not read it as 0.

How do I connect it?

To connect: paste this to your AI assistant: “Read https://dataxapi.com/agents.md and set up DataXAPI here. My API key is dx_…; use it for every DataXAPI call.” A client that takes a remote MCP server can be given https://api.dataxapi.com/mcp with the header Authorization: Bearer <key> instead.

What to ask once it is connected

Say what you want: the agent finds the endpoint and its price, then calls it; finding and reading parameters are free.

What does it cost?

Paid per call; a failed call is not charged. A page of 20 video search results is $0.0020; the free trial covers up to 333 calls. Give the agent its own key with a limit; X-Max-Price caps what one call may cost.

Why DataXAPI

• Steady: endpoints were available 100% of the last 30 days, on a public status page you can check.
• Fast: a paid call took 1.6 seconds at the median over the last 7 days (counted 2026-11-06).
• Fewer tokens: across 9 endpoints on Bilibili, one recorded real response each, the platform's own JSON comes to 477k tokens and what an AI reads through DataXAPI's MCP server to 21k, 96% fewer; the JSON of a direct API call is 95% fewer too (counted 2026-11-06). Every platform answers in one shape: the author in author, counts in metrics, the next page in page.next_cursor, so another platform needs no new parser.
• See before you pay: every endpoint's page shows a recorded real response, field by field.
• Spending you control: a failed call is not charged; each key can have a limit, X-Max-Price caps one call, and the same Idempotency-Key within 24 hours is never charged twice.
• Ready for AI: a remote MCP server listed in the official MCP registry as com.dataxapi/mcp, usable from Claude, Cursor and ChatGPT.
• Free trial of up to 333 calls, no card.

Endpoints

Each with its price; every page has the parameters, the fields and a recorded response.