Reads your question and decides whether it needs live information or can answer directly.
Calls web_search to find current, specific sources instead of answering from assumption.
Calls fetch_url to pull the full text of the most relevant pages before drawing conclusions.
Combines what it found into a clear, well-cited answer that references the source URLs.
For report-style requests, calls generate_pdf_report and links the downloadable PDF in its reply.
Connect your own LLM provider key in Settings. MicroManus never marks up or resells model usage.
Live web_search and fetch_url calls back real sources, not a static training-data guess.
Ask for a report and the agent assembles cited findings into a downloadable PDF artifact.
Every chat's token usage and dollar cost is broken down per message on the Stats page.
Sign in, add your key, and let MicroManus handle the rest.
$ authenticate --provider=google