Skip to content
Get packs

DocsGPT

Self-hosted platform for AI agents that answer from your own documents, with any model.

GitHub Download 1.1.0
Stars
18.4k
Forks
2.2k
Licence
MIT
Language
Python
Last update
4 days ago
Type
App

What it is

DocsGPT turns your documents, websites and connected apps into AI agents that answer questions and cite their sources. You can build agents and visual workflows, give them tools, and embed them in a website, an OpenAI-compatible API, or an MCP server. The README says it can run entirely in your own environment with cloud or local models.

  • Build an agent grounded in a company handbook and embed its chat widget on a support page
  • Connect a GitHub repository or Google Drive so the agent's knowledge stays in sync
  • Expose an agent through the OpenAI-compatible API so existing SDK code can call it
  • Run a scheduled agent or trigger one through a webhook
  • Let an MCP client such as Claude or Cursor search an agent's knowledge

Great for

  • Teams that want an internal assistant over their own documents and wikis
  • Developers who want to embed a documentation chatbot on a product site
  • People who need to keep data inside their own infrastructure
  • Users who want to run local models through Ollama or vLLM

Not ideal for

  • People who only want a quick hosted chatbot and no setup, since the cloud option is a separate service and the self-hosted route needs Docker and configuration
  • Non-technical users without someone to run the install and manage upgrades, backups and security settings
  • Teams that need guarantees about model quality, as the README does not make claims about answer accuracy beyond citing sources
Step-by-step guide: how to use DocsGPTInstall, first run, examples, tips and FAQs
DocsGPT alternativesDocsGPT vs VaneAnythingLLM vs DocsGPTDocsGPT vs LlamaGPTdoccano vs DocsGPT

How to get it

  1. 1

    See the README

    You'll need: Docker

    Install on macOS or Linux (one command, from README)

    curl -fsSL https://docs.ac/install | bash

    Install on Windows PowerShell (from README)

    irm https://docs.ac/install.ps1 | iex
  2. 2

    Run it

    • Run the installer command for your operating system
    • Answer the prompts from `docsgpt up` about who should reach DocsGPT and which model to use
    • Open http://localhost:7091 in your browser for a local install
    • Check the stack with `docsgpt status`
  3. 3

    Get the most out of it

    • Connect external sources such as Google Drive, GitHub or Confluence so agents stay current
    • Use visual workflows to chain agents with conditions
    • Embed an agent with the chat widget or React package on your own site
    • Use the OpenAI-compatible API or Agent API to call agents from your own code
    • Schedule `docsgpt backup` to protect the database and uploaded data

Downloads

Source code (.zip)main branch · straight from GitHubRelease 1.1.03 files incl. install.ps1Official website & docsapp.docsgpt.cloud/

Strengths

  • Self-hosting and the MIT license let you keep data in your environment
  • Supports many model providers, including local models
  • Answers include citations to the source documents

Trade-offs

  • Self-hosting involves several services (API, worker, Postgres, Redis, vector store), so it takes real setup effort
  • The README is truncated here, so detailed requirements and limits are not fully visible
  • Features like guardrails and deep research depend on configuration the README does not fully detail

How it compares

ProjectStarsTypeLevelUpdatedBest for
DocsGPT (this)18.4kAppintermediate4 days agoTeams that want an internal assistant over their own documents and wikis
anything-llm67.8kAppbeginner6 days agoPeople who want a private ChatGPT-style assistant over their own PDFs and text files
Prompt-Engineering-Guide79.8kLearning resourcebeginner8 months agoPeople who are new to prompting and want structured lessons with examples
Made-With-ML49.9kLearning resourceintermediate8 months agoSoftware engineers who want to learn how ML models get deployed and maintained
FinGPT21.5kAI model / agentadvanced12 days agoDevelopers who want to experiment with finance-tuned LLMs for sentiment analysis
Vane38kAppintermediate2 months agoPeople who want AI-written answers with links to sources

Stats synced from GitHub. The assessment above was drafted with AI from the project's README on 2026-10-29 and may miss recent changes — check the README before relying on a command. MIT licence: free to use, modify and distribute, including commercially, as long as you keep the copyright notice.