Practical overview
When it fits
Use it when web research needs an inspectable trail from report claims to source pages, or when a specific finding needs a follow-up search. It is also relevant when choosing a self-hosted research interface with configurable model and search providers.
Who it is for
- Analysts and researchers who need a cited first-pass report from web sources.
- Teams or individuals who want to review source excerpts and continue research on specific findings.
- Self-hosters choosing between browser-configured keys for Client Mode and server-side keys for Server Mode.
Documented uses
- Turn a research question into planned web searches and a cited report.
- Inspect the sources, retrieval details, and excerpts associated with report findings.
- Research a finding further and update related report passages while retaining the earlier version.
- Export a report as Markdown or PDF.
Requirements & limits
The project documents a live demo and self-hosting options, including Docker. Client Mode uses user-entered browser keys; Server Mode uses environment variables and requires an SSR/Nitro runtime. The public repository includes an MIT license.
- Client Mode suits static deployments and keeps configuration and API requests in the browser; Server Mode requires an SSR/Nitro runtime.
- The chosen AI and web-search providers need their corresponding configuration; the documentation lists multiple provider options.
- Follow-up research is limited to two directions per round and updates only Markdown blocks citing the finding.
- Research history is stored in the browser, so useful work should be exported for retention.
- A source-matched excerpt still requires contextual reading to assess the finding.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Review a report citation and follow up on a finding
Inspect a cited finding and create a follow-up report entry while retaining the original.
Before you start
- A completed research report in Deep Research Web UI.
- The application running in Client Mode or Server Mode.
Steps
Select a numbered citation in the report or choose “Inspect evidence and follow up.”
Enter the point to verify and choose “Follow up and update report.”
Review the updated report and its separate history entry; export research you want to keep.
Expected outcome
A successful follow-up saves new evidence and updates relevant cited Markdown blocks in a separate history entry, leaving the original report unchanged.
Guide limitations
- A follow-up searches at most two directions in one round.
- Failed, cancelled, or unmatched follow-ups preserve the original report.
Configure a self-hosted Server Mode deployment with Docker
Start the documented Docker image with server-side AI and web-search configuration.
Before you start
- Docker and an environment capable of running the Nuxt server.
- An AI provider API key and a web-search provider API key.
- Provider and model settings appropriate to the chosen services.
Steps
Set Server Mode and provide the AI and web-search keys as environment variables.
Set the AI provider, model, and web-search provider values for the configuration.
Run the documented Docker image with the variables and port mapping.
Expected outcome
The application runs in Server Mode with provider keys configured server-side, so users do not need to enter those keys in the UI.
Guide limitations
- Server Mode requires an SSR/Nitro runtime and is not available for purely static deployments.
- The example command uses a particular AI model and Tavily; provider-specific setup may differ.
- The documentation recommends an access password for publicly reachable Server Mode deployments.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 14 October 2026
- Original website title
- AnotiaWang/deep-research-web-ui · Official homepage title · retained in its source language