nilsherzig

LLocalSearch: Search the Web with Local LLM Agents

LLocalSearch runs a web-searching agent workflow with locally hosted language models, so users can ask questions without OpenAI or Google API keys. It is archived, so it is best suited to evaluation or self-managed use rather than an actively maintained deployment.

ArchivedGoApache-2.0LLMAI AgentsSelf Hosted
View on GitHub Compare
Stars
5.9k
Forks
365
Stars, last 30 days
-6
Last commit
7 months ago
Contributors
6

Our take

Archivedarchived by its owners on GitHub, read-only

Health score

20/100

How it is scored
Activity
19
Community
21
Issues
21
Pull requests
20

-6 stars in the last 30 days

LLocalSearch demonstrates an approachable local-agent search workflow, but its archived, inactive state and highly concentrated contributor base make it a poor choice for a deployment that needs ongoing maintenance.

Good fit if

  • You want to evaluate web search with local LLMs and can manage the software yourself.
  • Seeing agent progress and source links is useful to your research workflow.
  • You are comfortable with Docker Compose and the project’s Apache-2.0 license.

Look elsewhere if

  • You need an actively maintained project, timely fixes, or a release cadence.
  • Your deployment requires reliable support or ongoing issue and pull-request handling.
  • You need chat histories or user accounts, which the README lists as planned work rather than available features.
All health signals
Last commit2026-03-24 (7 months ago)
Commits, last 90 days0
Releases, last 12 months0 (latest v0.2, 2024-04-13)
Contributors6 (top contributor: 97% of commits)
Issues closed, last 90 days0
Pull requests merged, last 90 days0
Project age2 years

Checked on 2026-10-08 with the GitHub API.

Project history

Daily GitHub snapshots since Oct 7, 2026. History covers stars and forks; health signals are not tracked historically.

Selected snapshot

Stars

5,932

Forks

363

Change since Nov 15, 2026: -1 stars·0 forks

Overview

LLocalSearch is a self-hosted web app that uses locally running language models to investigate questions. An agent can call internet-search tools repeatedly, using results from earlier calls to guide further research, then present an answer with links and progress logs.

It is intended for people who want current web information without sending questions to hosted LLM APIs. The visible sources and agent activity can help users inspect how an answer was assembled, but do not guarantee that the answer is correct or complete.

Key Features

  • Runs with locally hosted language models, without OpenAI or Google API keys.
  • Lets agents use web-search tools recursively as they investigate a question.
  • Shows live agent logs and links to sources alongside answers.
  • Supports follow-up questions.
  • Provides a mobile-friendly interface with dark and light modes.
  • Uses Docker Compose for setup and deployment.

Use Cases

  • Privacy-conscious individuals can research current topics without using hosted LLM APIs, provided they are comfortable running the local model and app.
  • Researchers can use the agent to gather web sources and inspect the links behind a generated answer as a starting point for further reading.
  • Developers can explore a locally hosted, tool-using LLM search workflow or adapt the project, with the caveat that it is archived.

What you need

Detected in the repository

  • Node.js (from package.json)
  • A Dockerfile, so it can run in a container; a Compose file sets up the related services

License in plain words

Apache-2.0permissive

  • Commercial use: yes
  • Modify and redistribute: yes
  • You must keep: the license, the NOTICE file and a note of your changes
  • Share your changes: no
  • Includes an explicit patent grant from the contributors:

A summary, not legal advice: the LICENSE file is what applies.

Getting Started

Clone the repository and start the containers:

git clone https://github.com/nilsherzig/LLocalSearch.git
cd LLocalSearch
docker-compose up -d

See the README and Ollama setup guide for configuration details.

Alternatives

  • local-deep-research: It produces cited reports using web, academic, and private-document search, rather than focusing specifically on web search with local models.
  • AingDesk: It combines web search with knowledge bases and a broader assistant interface, and can use cloud as well as local models.
  • lemonai: It combines research browsing with coding and data analysis in a self-hostable agent with a VM sandbox.
ProjectLanguageLicenseStarsStatus
LLocalSearchGoApache-2.05.9kArchived
local-deep-researchPythonMIT9.2kActive
AingDeskTypeScriptMIT2.5kQuiet
lemonaiJavaScriptOther1.6kQuiet

Considerations

The repository is archived and has had no recent commits, releases, or issue closures. There has been no release in over a year, and most of the code comes from one person, so adopters should expect to handle maintenance and compatibility work themselves. The repository reports six contributors overall, but recent pull-request and issue activity is absent.

The app is designed to work with locally hosted LLMs, and its setup uses Docker Compose. The README describes running on low-end hardware, but local model performance will depend on the chosen model and available hardware. It also identifies chat histories and user accounts as unfinished plans. The project uses Apache-2.0; review the license terms before redistributing or modifying it.

Found this useful?

Share it with someone who would like LLocalSearch.

Comparisons