Two ways to see the package in action: a browser app and a CLI. Each talks to the live Diffbot APIs, and both use an Anthropic model to drive their agent, so you need a DIFFBOT_API_TOKEN and an ANTHROPIC_API_KEY. Copy .env.example to .env and fill it in.

Every example has its own README covering setup and how to run it.

Each example resolves @diffbot/langchain through link:../.., which points at the package's build output — so run pnpm build in the repo root once before installing an example.

dql-explorer/ is a browser UI over the Knowledge Graph with three tabs:

• DQL Builder — type a question in plain English; an agent inspects the ontology, probes query variants for selectivity, and commits to a DQL query. The agent only authors the query — the server runs it — so the rows in the table are always real Knowledge Graph data, never model output.
• M&A / IPO Dashboard — a deterministic roll-up of two fixed DQL templates into charts by industry, month, country, and exchange. No model involved.
• Ask Diffbot — ChatDiffbot streaming a RAG answer token by token.

A Hono backend serving a React + TypeScript frontend on a single port.

company-research/ is a multi-tool agent packaged as a one-shot command-line tool: ask a company-research question in plain English and the agent searches the Knowledge Graph and the live web, then cites the entity IDs and URLs it used. Useful for shell scripting or quick spot checks.

The Python package ships a quickstart/ Jupyter notebook as its guided tour. There is no equivalent here — notebooks are not idiomatic in the TypeScript ecosystem, and the material it covers (retrievers, output shaping, the tools, ChatDiffbot streaming, bring-your-own client) is covered by runnable snippets in the root README, every one of which is executed by the test suite on each run.