# Technical Documentation: khoj-ai/khoj > ℹ️ **Provenance:** Multi-Tier Ingestion Β· [CodeWiki Reference](https://codewiki.google/github.com/khoj-ai/khoj) ### Overview (README.md)

Khoj Logo

[](https://github.com/khoj-ai/khoj/actions/workflows/test.yml) [](https://github.com/khoj-ai/khoj/pkgs/container/khoj) [](https://pypi.org/project/khoj/) [](https://discord.gg/BDgyabRM6e)
Your AI second brain

[πŸ“‘ Docs](https://docs.khoj.dev)   β€’   [🌐 Web](https://khoj.dev)   β€’   [πŸ”₯ App](https://app.khoj.dev)   β€’   [πŸ’¬ Discord](https://discord.gg/BDgyabRM6e)   β€’   [✍🏽 Blog](https://blog.khoj.dev) khoj-ai%2Fkhoj | Trendshift
*** ### 🎁 New * Meet 🌢️ **[Pipali](https://pipali.ai)** - our [open-source](https://github.com/khoj-ai/pipali) AI coworker that runs on your computer. * [Read](https://blog.khoj.dev/posts/evaluate-khoj-quality/) about Khoj's excellent performance on modern retrieval and reasoning benchmarks. *** ## Overview [Khoj](https://khoj.dev) is a personal AI app to extend your capabilities. It smoothly scales up from an on-device personal AI to a cloud-scale enterprise AI. - Chat with any local or online LLM (e.g llama3, qwen, gemma, mistral, gpt, claude, gemini, deepseek). - Get answers from the internet and your docs (including image, pdf, markdown, org-mode, word, notion files). - Access it from your Browser, Obsidian, Emacs, Desktop, Phone or Whatsapp. - Create agents with custom knowledge, persona, chat model and tools to take on any role. - Automate away repetitive research. Get personal newsletters and smart notifications delivered to your inbox. - Find relevant docs quickly and easily using our advanced semantic search. - Generate images, talk out loud, play your messages. - Khoj is open-source, self-hostable. Always. - Run it privately on [your computer](https://docs.khoj.dev/get-started/setup) or try it on our [cloud app](https://app.khoj.dev). *** ## See it in action Go to https://app.khoj.dev to see Khoj live. ## Full feature list You can see the full feature list [here](https://docs.khoj.dev/category/features). ## Self-Host To get started with self-hosting Khoj, [read the docs](https://docs.khoj.dev/get-started/setup). ## Enterprise Khoj is available as a cloud service, on-premises, or as a hybrid solution. To learn more about Khoj Enterprise, [visit our website](https://khoj.dev/teams). ## Frequently Asked Questions (FAQ) Q: Can I use Khoj without self-hosting? Yes! You can use Khoj right away at [https://app.khoj.dev](https://app.khoj.dev) β€” no setup required. Q: What kinds of documents can Khoj read? Khoj supports a wide variety: PDFs, Markdown, Notion, Word docs, org-mode files, and more. Q: How can I make my own agent? Check out [this blog post](https://blog.khoj.dev/posts/create-agents-on-khoj/) for a step-by-step guide to custom agents. For more questions, head over to our [Discord](https://discord.gg/BDgyabRM6e)! ## Contributors Cheers to our awesome contributors! πŸŽ‰ Made with [contrib.rocks](https://contrib.rocks). ### Interested in Contributing? Khoj is open source. It is sustained by the community and we’d love for you to join it! Whether you’re a coder, designer, writer, or enthusiast, there’s a place for you. Why Contribute? - Make an Impact: Help build, test and improve a tool used by thousands to boost productivity. - Learn & Grow: Work on cutting-edge AI, LLMs, and semantic search technologies. You can help us build new features, improve the project documentation, report issues and fix bugs. If you're a developer, please see our [Contributing Guidelines](https://docs.khoj.dev/contributing/development) and check out [good first issues](https://github.com/khoj-ai/khoj/contribute) to work on. --- ### Documentation README (`documentation/README.md`) # Website This website is built using [Docusaurus](https://docusaurus.io/), a modern static website generator. ### Installation ``` $ yarn ``` ### Local Development ``` $ yarn start ``` This command starts a local development server and opens up a browser window. Most changes are reflected live without having to restart the server. ### Build ``` $ yarn build ``` This command generates static content into the `build` directory and can be served using any static contents hosting service. ### Deployment Using SSH: ``` $ USE_SSH=true yarn deploy ``` Not using SSH: ``` $ GIT_USER= yarn deploy ``` If you are using GitHub pages for hosting, this command is a convenient way to build the website and push to the `gh-pages` branch. --- ### Documentation Docs Get Started Overview (`documentation/docs/get-started/overview.md`) --- sidebar_position: 0 slug: / keywords: ["khoj", "khoj ai", "khoj docs", "khoj documentation", "khoj features", "khoj overview", "khoj quickstart", "khoj chat", "khoj search", "khoj cloud", "khoj self-host", "khoj setup", "open source ai", "local llm", "ai copilot", "second brain", "personal ai", "ai search engine", "research assistant", "open source llm", "self-host llm", "self-host chatgpt", "ai chatbot", "ai assistant", "ai research assistant", "ai search engine", "ai knowledge base", "ai knowledge graph", "ai personal assistant", "ai second brain", "open source ai assistant"] --- # Overview
Your Second Brain

[πŸ“œ Explore Code](https://github.com/khoj-ai/khoj)   β€’   [🌍 Try Khoj Cloud](https://khoj.dev)   β€’   [πŸ’¬ Get Involved](https://discord.gg/BDgyabRM6e)
## Introduction Welcome to the Khoj Docs! This is the best place to get setup and explore Khoj's features. - Khoj is an open source, personal AI - You can [chat](/features/chat) with it about anything. It'll use files you shared with it to respond, when relevant. It can also access information from the public internet. - Quickly [find](/features/search) relevant notes and documents using natural language - It understands pdf, plaintext, markdown, org-mode files, and [notion pages](/data-sources/notion_integration). - Access it from your [Emacs](/clients/emacs), [Obsidian](/clients/obsidian), the [Khoj desktop app](/clients/desktop), or [any web browser](/clients/web) - Use our [cloud](https://app.khoj.dev) instance to access your Khoj anytime from anywhere, [self-host](/get-started/setup) on consumer hardware for privacy ## Quickstart - [Try Khoj Cloud](https://app.khoj.dev) to get started quickly - [Read these instructions](/get-started/setup) to self-host a private instance of Khoj --- ### Documentation Docs Advanced Admin (`documentation/docs/advanced/admin.md`) # Admin Panel > Describes the Khoj settings configurable via the admin panel By default, your admin panel is available at `http://localhost:42110/server/admin/`. You can access the admin panel by logging in with your admin credentials (this would be your `KHOJ_ADMIN_EMAIL` and `KHOJ_ADMIN_PASSWORD`). The admin panel allows you to configure various settings for your Khoj server. ## App Settings ### Agents Add all the agents you want to use for your different use-cases like Writer, Researcher, Therapist etc. - `Personality`: This is a prompt to tell the chat model how to tune the personality of the agent. - `Chat model`: The chat model to use for the agent. - `Name`: The name of the agent. This field helps give the agent a unique identity across the app. - `Avatar`: Url to the agents profile picture. It helps give the agent a unique visual identity across the app. - `Style color`, `Style icon`: These fields help give the agent a unique, visually identifiable identity across the app. - `Slug`: This is the agent name to use in urls. - `Public`: Check this if the agent is expected to be visible to all users on this Khoj server. - `Managed by admin`: Check this if the agent is managed by admin, not by any user. - `Creator`: The user who created the agent. - `Tools`: The list of tools available to this agent. Tools include notes, image, online. This field is not currently configurable and only supports all tools (i.e `["*"]`) ### Chat Model Options Add all the chat models you want to try, use and switch between for your different use-cases. For each chat model you add: - `Chat model`: The name of an [OpenAI](https://platform.openai.com/docs/models), [Anthropic](https://docs.anthropic.com/en/docs/about-claude/models#model-names), [Gemini](https://cloud.google.com/vertex-ai/generative-ai/docs/learn/models#gemini-models) or [Offline](https://huggingface.co/models?pipeline_tag=text-generation&library=gguf) chat model. - `Model type`: The chat model provider like `OpenAI`, `Google`. - `Vision enabled`: Set to `true` if your model supports vision. This is currently only supported for vision capable OpenAI models like `gpt-4o` - `Max prompt size`, `Subscribed max prompt size`: These are optional fields. They are used to truncate the context to the maximum context size that can be passed to the model. This can help with accuracy and cost-saving.
- `Tokenizer`: This is an optional field. It is used to accurately count tokens and truncate context passed to the chat model to stay within the models max prompt size. ### Server Chat Settings The server chat settings are used as: 1. The default chat models for subscribed (`Advanced` field) and unsubscribed (`Default` field) users. 2. The chat model for all intermediate steps like intent detection, web search etc. during chat response generation. If a server chat setting is not added the first ChatModelOption in your config is used as the default chat model. To add a server chat setting: - Set your preferred default chat models in the `Default` fields of your [ServerChatSettings](http://localhost:42110/server/admin/database/serverchatsettings/) - The `Advanced` field doesn't need to be set when self-hosting. When unset, the `Default` chat model is used for all users and the intermediate steps. ### AI Model API These settings configure APIs to interact with AI models. For each AI Model API you [add](http://localhost:42110/server/admin/database/aimodelapi/add): - `Api key`: Set to your [OpenAI](https://platform.openai.com/api-keys), [Anthropic](https://console.anthropic.com/account/keys) or [Gemini](https://aistudio.google.com/app/apikey) API keys. - `Name`: Give the configuration any friendly name like `OpenAI`, `Gemini`, `Anthropic`. - `Api base url`: Set the API base URL. This is only relevant to set if you're using another OpenAI-compatible proxy server like [Ollama](/advanced/ollama) or [LMStudio](/advanced/lmstudio). ### Search Model Configs Search models are used to generate vector embeddings of your documents for natural language search and chat. You can choose any [embeddings models on HuggingFace](https://huggingface.co/models?pipeline_tag=sentence-similarity) to create vector embeddings of your documents for natural language search and chat. Example Search Model Settings ### Text to Image Model Options Add text to image generation models with these settings. Khoj currently supports text to image models available via OpenAI, Google or Replicate API - `api-key`: Set to your OpenAI, Google AI or Replicate API key - `model`: Set the model name available over the selected model provider - `model-type`: Set to the appropriate model provider - `openai-config`: For image generation models available via OpenAI (compatible) API you can set the appropriate OpenAI Processor Conversation Settings instead of specifying the `api-key` field above ### Speech to Text Model Options Add speech to text models with these settings. Khoj currently only supports whisper speech to text model via OpenAI API or Offline ### Voice Model Options Add text to speech models with these settings. Khoj currently supports models from [ElevenLabs](https://elevenlabs.io/). ### Reflective Questions This is a static list of starter question suggestions for each user. It is not currently used in any client app. It used to be shown on the web app home page. We may turn it into a dynamic list of starter questions personalized to each users, say based on their recent conversations or synced knowledge base. ## User Data - Users, Entrys, Conversations, Subscriptions, Github configs, Notion configs, User search configs, User conversation configs, User voice configs ## Miscellaneous Data - Process Locks: Persistent Locks for Automations - Client Applications: Client applications allow you to setup third party applications that can query your Khoj server using a client application ID + secret. The secret would go in a bearer token. --- ### Documentation Docs Advanced Authentication (`documentation/docs/advanced/authentication.mdx`) # Authenticate (Multi-User Setup) ```mdx-code-block import Tabs from '@theme/Tabs'; import TabItem from '@theme/TabItem'; ``` By default, most of the instructions for self-hosting Khoj assume a single user, and so the default configuration is to run in anonymous mode. However, if you want to enable authentication, you can do so either with with [Magic Links](#using-magic-links) or [Google OAuth](#using-google-oauth) as shown below. This can be helpful to make Khoj securely accessible to you and your team. :::tip[Note] Remove the `--anonymous-mode` flag from your khoj start up command or docker-compose file to enable authentication. ::: ## Using Magic Links The most secure way to do this is to integrate with [Resend](https://resend.com). 1. Setup your account at https://resend.com 2. Set an environment variable for `RESEND_API_KEY`. You can get your API key [here](https://resend.com/api-keys). 3. Set an environment variable for `RESEND_EMAIL`. This is the email address that will show up in your `from` field when sending magic links. This will allow you to automatically send sign-in links to users who want to log in. It's still possible to use the magic links feature without Resend, but you'll need to manually send the magic links to users who want to log in. ## Manually sending magic links 1. The user will have to enter their email address in the login popup shown at http://localhost:42110/?v=app. They'll click `Get Login Link`. Without the Resend API key, this will just create an unverified account for them in the backend Magic link login form 2. You can get their magic link using the admin panel Go to the [admin panel](http://localhost:42110/server/admin/database/khojuser/). You'll see a list of users. Search for the user you want to send a magic link to. Tick the checkbox next to their row, and use the action drop down at the top to 'Get email login URL'. This will generate a magic link that you can send to the user, which will appear at the top of the admin interface. | Get email login URL | Retrieved login URL | |---------------------|---------------------| | Get user magic sign in link| Successfully retrieved a login URL| 3. Send the magic link to the user. They can click on it to log in. Once they click on the link, they'll automatically be logged in. They'll have to repeat this process for every new device they want to log in from, but they shouldn't have to repeat it on the same device. A given magic link can only be used once. If the user tries to use it again, they'll be redirected to the login page to get a new magic link. ## Using Google OAuth For this method, you'll need to use the prod version of the Khoj package. You can install it as below: Update your `docker-compose.yml` to use the prod image ```bash image: ghcr.io/khoj-ai/khoj-cloud:latest ``` ```bash pip install khoj[prod] ``` To set up your self-hosted Khoj with Google Auth, you need to create a project in the Google Cloud Console and enable the Google Auth API. To implement this, you'll need to: 1. [Create authorization credentials](https://developers.google.com/identity/sign-in/web/sign-in) for your application. 2. Open your [Google cloud console](https://console.developers.google.com/apis/credentials) and create a configuration like below for the relevant `OAuth 2.0 Client IDs` project: 3. Configure these environment variables: `GOOGLE_CLIENT_SECRET`, and `GOOGLE_CLIENT_ID`. You can find these values in the Google cloud console, in the same place where you configured the authorized origins and redirect URIs. That's it! That should be all you have to do. Now, when you reload Khoj without `--anonymous-mode`, you should be able to use your Google account to sign in. --- ### Documentation Docs Advanced Gcp Vertex (`documentation/docs/advanced/gcp-vertex.md`) # Google Vertex AI :::info This is only helpful for self-hosted users. If you're using [Khoj Cloud](https://app.khoj.dev), you can directly use any of the pre-configured AI models. ::: Khoj can use Google's Gemini and Anthropic's Claude family of AI models from [Vertex AI](https://cloud.google.com/vertex-ai) on Google Cloud. Explore Anthropic and Gemini AI models available on Vertex AI's [Model Garden](https://console.cloud.google.com/vertex-ai/model-garden). ## Setup 1. Follow [these instructions](https://cloud.google.com/vertex-ai/generative-ai/docs/partner-models/use-claude#before_you_begin) to use models on GCP Vertex AI. 2. Create [Service Account](https://console.cloud.google.com/apis/credentials/serviceaccountkey) credentials. - Download the credentials keyfile in json format. - Base64 encode the credentials json keyfile. For example by running the following command from your terminal: `base64 -i ` 3. Create a new [API Model API](http://localhost:42110/server/admin/database/aimodelapi/add) on your Khoj admin panel. - **Name**: `Google Vertex` (or whatever friendly name you prefer). - **Api Key**: `base64 encoded json keyfile` from step 2. - **Api Base Url**: `https://{MODEL_GCP_REGION}-aiplatform.googleapis.com/v1/projects/{YOUR_GCP_PROJECT_ID}` - MODEL_GCP_REGION: A region the AI model is available in. For example `us-east5` works for [Claude](https://cloud.google.com/vertex-ai/generative-ai/docs/partner-models/use-claude#regions). - YOUR_GCP_PROJECT_ID: Get your project id from the [Google cloud dashboard](https://console.cloud.google.com/home/dashboard) 4. Create a new [Chat Model](http://localhost:42110/server/admin/database/chatmodel/add) on your Khoj admin panel. - **Name**: `claude-3-7-sonnet@20250219`. Any Claude or Gemini model on Vertex's Model Garden should work. - **Model Type**: `Anthropic` or `Google` - **Ai Model API**: *the Google Vertex Ai Model API you created in step 3* - **Max prompt size**: `60000` (replace with the max prompt size of your model) - **Tokenizer**: *Do not set* 5. Select the chat model on [your settings page](http://localhost:42110/settings) and start a conversation. ## Troubleshooting & gcp AI Tips - Permission Denied? Ensure your service account has the `Vertex AI User` role and that the API is enabled in your GCP project. - Region Errors? Double-check that the model you're trying to use is supported in your selected region. Some Claude or Gemini models are restricted to specific zones like `us-east5` or `us-central1`. - Prompt Size Limitations The "Max prompt size" should align with the limits defined in the model documentation. Exceeding it can silently fail or truncate inputs. - Testing the API Key Before adding it to Khoj, you can verify that your key works by making a simple curl request to Vertex AI. This helps debug auth issues early. - Use Environment Variables For better security, consider using environment variables to manage sensitive keys and inject them at runtime during base64 encoding. If you encounter any issues, the [Khoj Discord](https://discord.gg/BDgyabRM6e) is a great place to ask for help! --- ### Documentation Docs Advanced Litellm (`documentation/docs/advanced/litellm.md`) # LiteLLM :::info This is only helpful for self-hosted users. If you're using [Khoj Cloud](https://app.khoj.dev), you're limited to our first-party models. ::: :::info Khoj natively supports local LLMs [available on HuggingFace in GGUF format](https://huggingface.co/models?library=gguf). Using an OpenAI API proxy with Khoj maybe useful for ease of setup, trying new models or using commercial LLMs via API. ::: [LiteLLM](https://docs.litellm.ai/docs/proxy/quick_start) exposes an OpenAI compatible API that proxies requests to other LLM API services. This provides a standardized API to interact with both open-source and commercial LLMs. Using LiteLLM with Khoj makes it possible to turn any LLM behind an API into your personal AI agent. ## Setup 1. Install LiteLLM ```bash pip install litellm[proxy] ``` 2. Start LiteLLM and use Mistral tiny via Mistral API ``` export MISTRAL_API_KEY= litellm --model mistral/mistral-tiny --drop_params ``` 3. Create a new [API Model API](http://localhost:42110/server/admin/database/aimodelapi/add) on your Khoj admin panel - **Name**: `litellm` - **Api Key**: `any string` - **Api Base Url**: `` 4. Create a new [Chat Model](http://localhost:42110/server/admin/database/chatmodel/add) on your Khoj admin panel. - **Name**: `llama3.1` (replace with the name of your local model) - **Model Type**: `Openai` - **Ai Model Api**: *the litellm Ai Model API you created in step 3* - **Max prompt size**: `20000` (replace with the max prompt size of your model) - **Tokenizer**: *Do not set for OpenAI, Mistral, Llama3 based models* 5. Go to [your config](http://localhost:42110/settings) and select the model you just created in the chat model dropdown. --- ### Documentation Docs Advanced Lmstudio (`documentation/docs/advanced/lmstudio.md`) # LM Studio :::warning[Unsupported] Khoj does not work with LM Studio anymore. Khoj leverages [json mode](https://platform.openai.com/docs/guides/structured-outputs#json-mode) extensively but LMStudio's API seems to have dropped support for json mode. [1](https://x.com/lmstudio/status/1770135858709975547), [2](https://lmstudio.ai/docs/api/structured-output) ::: :::info This is only helpful for self-hosted users. If you're using [Khoj Cloud](https://app.khoj.dev), you're limited to our first-party models. ::: :::info Khoj natively supports local LLMs [available on HuggingFace in GGUF format](https://huggingface.co/models?library=gguf). Using an OpenAI API proxy with Khoj maybe useful for ease of setup, trying new models or using commercial LLMs via API. ::: [LM Studio](https://lmstudio.ai/) is a desktop app to chat with open-source LLMs on your local machine. LM Studio provides a neat interface for folks comfortable with a GUI. LM Studio can expose an [OpenAI API compatible server](https://lmstudio.ai/docs/local-server). This makes it possible to turn chat models from LM Studio into your personal AI agents with Khoj. ## Setup 1. Install [LM Studio](https://lmstudio.ai/) and download your preferred Chat Model 2. Go to the Server Tab on LM Studio, Select your preferred Chat Model and Click the green Start Server button 3. Create a new [AI Model API](http://localhost:42110/server/admin/database/aimodelapi/add/) on your Khoj admin panel - **Name**: `lmstudio` - **Api Key**: `any string` - **Api Base Url**: `http://localhost:1234/v1/` (default for LMStudio) 4. Create a new [Chat Model](http://localhost:42110/server/admin/database/chatmodel/add) on your Khoj admin panel. - **Name**: `llama3.1` (replace with the name of your local model) - **Model Type**: `Openai` - **Ai Model Api**: *the lmstudio Ai Model Api you created in step 3* - **Max prompt size**: `20000` (replace with the max prompt size of your model) - **Tokenizer**: *Do not set for OpenAI, mistral, llama3 based models* 5. Go to [your config](http://localhost:42110/settings) and select the model you just created in the chat model dropdown. --- ### Documentation Docs Advanced Ollama (`documentation/docs/advanced/ollama.mdx`) # Ollama ```mdx-code-block import Tabs from '@theme/Tabs'; import TabItem from '@theme/TabItem'; ``` :::info This is only helpful for self-hosted users. If you're using [Khoj Cloud](https://app.khoj.dev), you can use our first-party supported models. ::: :::info Khoj can directly run local LLMs [available on HuggingFace in GGUF format](https://huggingface.co/models?library=gguf). The integration with Ollama is useful to run Khoj on Docker and have the chat models use your GPU or to try new models via CLI. ::: Ollama allows you to run [many popular open-source LLMs](https://ollama.com/library) locally from your terminal. For folks comfortable with the terminal, Ollama's terminal based flows can ease setup and management of chat models. Ollama exposes a local [OpenAI API compatible server](https://github.com/ollama/ollama/blob/main/docs/openai.md#models). This makes it possible to use chat models from Ollama with Khoj. ## Setup :::info Restart your Khoj server after first run or update to the settings below to ensure all settings are applied correctly. ::: 1. Setup Ollama: https://ollama.com/ 2. Download your preferred chat model with Ollama. For example, ```bash ollama pull llama3.1 ``` 3. Uncomment `OPENAI_BASE_URL` environment variable in your downloaded Khoj [docker-compose.yml](https://github.com/khoj-ai/khoj/blob/master/docker-compose.yml#:~:text=OPENAI_BASE_URL) 4. Start Khoj docker for the first time to automatically integrate and load models from the Ollama running on your host machine ```bash # run below command in the directory where you downloaded the Khoj docker-compose.yml docker-compose up ``` 1. Setup Ollama: https://ollama.com/ 2. Download your preferred chat model with Ollama. For example, ```bash ollama pull llama3.1 ``` 3. Set `OPENAI_BASE_URL` environment variable to `http://localhost:11434/v1/` in your shell before starting Khoj for the first time ```bash export OPENAI_BASE_URL="http://localhost:11434/v1/" khoj --anonymous-mode ``` 1. Setup Ollama: https://ollama.com/ 2. Download your preferred chat model with Ollama. For example, ```bash ollama pull llama3.1 ``` 3. Create a new [AI Model API](http://localhost:42110/server/admin/database/aimodelapi/add) on your Khoj admin panel - **Name**: `ollama` - **Api Key**: `any string` - **Api Base Url**: `http://localhost:11434/v1/` (default for Ollama) 4. Create a new [Chat Model](http://localhost:42110/server/admin/database/chatmodel/add) on your Khoj admin panel. - **Name**: `llama3.1` (replace with the name of your local model) - **Model Type**: `Openai` - **AI Model API**: *the ollama AI Model API you created in step 3* - **Max prompt size**: `20000` (replace with the max prompt size of your model) 5. Go to [your config](http://localhost:42110/settings) and select the model you just created in the chat model dropdown. If you want to add additional models running on Ollama, repeat step 4 for each model. That's it! You should now be able to chat with your Ollama model from Khoj. --- ### Documentation Docs Advanced Remote (`documentation/docs/advanced/remote.md`) # Remote Access By default self-hosted Khoj is only accessible on the machine it is running. To securely access it from a remote machine: - Set the `KHOJ_DOMAIN` environment variable to your remotely accessible ip or domain via shell or docker-compose.yml. Examples: `KHOJ_DOMAIN=my.khoj-domain.com`, `KHOJ_DOMAIN=192.168.0.4`. - Ensure the Khoj Admin password and `KHOJ_DJANGO_SECRET_KEY` environment variable are securely set. - Setup [Authentication](/advanced/authentication). - Open access to the Khoj port (default: 42110) from your OS and Network firewall. :::warning[Use HTTPS certificate] To expose Khoj on a custom domain over the public internet, use of an SSL certificate is strongly recommended. You can use [Let's Encrypt](https://letsencrypt.org/) to get a free SSL certificate for your domain. To disable HTTPS, set the `KHOJ_NO_HTTPS` environment variable to `True`. This can be useful if Khoj is only accessible behind a secure, private network. ::: :::info[Try Tailscale] You can use [Tailscale](https://tailscale.com/) for easy, secure access to your self-hosted Khoj over the network. 1. Set `KHOJ_DOMAIN` to your machines [tailscale ip](https://tailscale.com/kb/1452/connect-to-devices#identify-your-devices) or [fqdn on tailnet](https://tailscale.com/kb/1081/magicdns#fully-qualified-domain-names-vs-machine-names). E.g `KHOJ_DOMAIN=100.4.2.0` or `KHOJ_DOMAIN=khoj.tailfe8c.ts.net` 2. Access Khoj by opening `http://tailscale-ip-of-server:42110` or `http://fqdn-of-server:42110` from any device on your tailscale network ::: --- ### Documentation Docs Advanced Support Multilingual Docs (`documentation/docs/advanced/support-multilingual-docs.md`) # Support Multilingual Docs Khoj uses an embedding model to understand documents. Multilingual embedding models improve the search quality for documents not in English. This affects both search and chat with docs experiences across Khoj. To improve search and chat quality for non-english documents you can use a [multilingual model](https://www.sbert.net/docs/pretrained_models.html#multi-lingual-models).
For example, the [paraphrase-multilingual-MiniLM-L12-v2](https://huggingface.co/sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2) supports [50+ languages](https://www.sbert.net/docs/pretrained_models.html#:~:text=we%20used%20the%20following%2050%2B%20languages), has decent search quality and speed for a consumer machine. To use it: 1. Open [the search config](http://localhost:42110/server/admin/database/searchmodelconfig/) on your server's admin settings page. Either create a new search model, if none exists, or update the existing one. For example, - Set the `bi_encoder` field to `sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2` - Set the `cross_encoder` field to `mixedbread-ai/mxbai-rerank-xsmall-v1` 2. Regenerate your content index from all the relevant clients. This step is very important, as you'll need to re-encode all your content with the new model. :::info[Note] Modern search/embedding model like [mixedbread-ai/mxbai-embed-large-v1](https://huggingface.co/mixedbread-ai/mxbai-embed-large-v1) expect a prefix to the query (or docs) string to improve encoding. Update the `bi_encoder_query_encode_config` field of your [embedding model](http://localhost:42110/server/admin/database/searchmodelconfig/) with `{prompt: }` to improve the search quality of these models. E.g. `{prompt: "Represent this query for searching documents"}`. You can pass any valid JSON object that the SentenceTransformer `encode` function accepts ::: --- ### Documentation Docs Advanced Tailscale (`documentation/docs/advanced/tailscale.md`) # Tailscale :::info This is only helpful for secure cross-device access to **self-hosted** Khoj. You **do not** need this if you're using [Khoj Cloud](https://app.khoj.dev). ::: [Tailscale](https://tailscale.com) simplifies creating a private VPN using [Wireguard](https://www.wireguard.com/) and OAuth. So you can host and access services on your devices from anywhere. The instructions below are one way to simply and securely access your self-hosted Khoj from your phone, laptop etc. ### Minimal Setup 1. Setup khoj on your preferred machine following the [standard steps](/get-started/setup) 2. Sign-up to [Tailscale](https://tailscale.com) and install the app on machines you want to access Khoj from. This usually includes your khoj server, your phone, laptop. Note the tailscale i.p of your khoj server. 3. Start khoj on your server by including the flag `--host ` 4. Open `http://:42110` to access khoj from any device on your tailscale network! ### HTTPS Certificate :::info Tailscale uses Wireguard to encrypt and route traffic between your machines. So HTTPS isn't required with Tailscale for secure access. HTTPS with Tailscale is only useful for browsers to not complain about security and block certain features like clipboard access unless HTTPS is enabled. ::: 1. Enable [MagicDNS](https://tailscale.com/kb/1081/magicdns#enabling-magicdns) and [HTTPS](https://tailscale.com/kb/1153/enabling-https) toggle on your tailscale admin console [DNS](https://login.tailscale.com/admin/dns) page. Note your unique tailscale domain name (usually ends with .ts.net) 2. Create an https certificate for your Khoj server by running the following command: ```bash # Assuming the server is named, `server` and your tailnet is `black-forest.ts.net` # Note path of the .crt and .key files generated tailscale cert server.black-forest.ts.net ``` 3. Start khoj to be served via https on standard port ```bash sudo KHOJ_DOMAIN=server.black-forest.ts.net \ khoj \ --sslcert /path/to/your/tailscale.crt \ --sslkey path/to/your/tailscale.key \ --host=server.black-forest.ts.net \ --port 443 ``` 4. You should now be able to access khoj on `https://server.black-forest.ts.net` from any device on your private tailscale network! --- ### Documentation Docs Advanced Use Openai Proxy (`documentation/docs/advanced/use-openai-proxy.md`) --- sidebar_position: 1 --- # Use OpenAI Proxy :::info This is only helpful for self-hosted users. If you're using [Khoj Cloud](https://app.khoj.dev), you're limited to our first-party models. ::: :::info Khoj natively supports local LLMs [available on HuggingFace in GGUF format](https://huggingface.co/models?library=gguf). Using an OpenAI API proxy with Khoj maybe useful for ease of setup, trying new models or using commercial LLMs via API. ::: Khoj can use any OpenAI API compatible server including local providers like [Ollama](/advanced/ollama), [LMStudio](/advanced/lmstudio) and [LiteLLM](/advanced/litellm) and commercial providers like [HuggingFace](https://huggingface.co/docs/api-inference/tasks/chat-completion#using-the-api), [OpenRouter](https://openrouter.ai/docs/quick-start) etc. Configuring this allows you to use non-standard, open or commercial, local or hosted LLM models for Khoj. Combine them with Khoj can turn your favorite LLM into an AI agent. Allowing you to chat with your docs, find answers from the internet, build custom agents and run automations. For specific integrations, see our [Ollama](/advanced/ollama), [LMStudio](/advanced/lmstudio) and [LiteLLM](/advanced/litellm) setup docs. For general instructions to setup Khoj with an OpenAI API proxy see below. ## General Setup 1. Start your preferred OpenAI API compatible app locally or get API keys from commercial AI model providers. 3. Create a new [API Model API](http://localhost:42110/server/admin/database/aimodelapi/add) on your Khoj admin panel - **Name**: `any name` - **Api Key**: `any string` - **Api Base Url**: *The URL of your Openai Compatible API* 3. Create a new [Chat Model](http://localhost:42110/server/admin/database/chatmodel/add) on your Khoj admin panel. - **Name**: `llama3` (replace with the name of your local model) - **Model Type**: `Openai` - **Ai Model Api**: *The AI Model API you created in step 2* - **Max prompt size**: `2000` (replace with the max prompt size of your model) - **Tokenizer**: *Do not set for OpenAI, mistral, llama3 based models* 4. Go to [your config](http://localhost:42110/settings) and select the model you just created in the chat model dropdown. --- ### Documentation Docs Clients Desktop (`documentation/docs/clients/desktop.md`) --- sidebar_position: 1 --- # Desktop > Upload your knowledge base to Khoj and chat with your whole corpus ## Companion App Share your files, folders with Khoj using the app. Khoj will keep these files in sync to provide contextual responses when you search or chat. ## Setup :::info[Self Hosting] If you are self-hosting the Khoj server, update the *Settings* page on the Khoj Desktop app to: - Set the `Khoj URL` field to your Khoj server URL. By default, use `http://127.0.0.1:42110`. - Do not set the `Khoj API Key` field if your Khoj server runs in anonymous mode. For example, `khoj --anonymous-mode` ::: 1. Install the [Khoj Desktop app](https://khoj.dev/downloads) for your OS 2. Generate an API key on the [Khoj Web App](https://app.khoj.dev/settings#clients) 3. Set your Khoj API Key on the *Settings* page of the Khoj Desktop app 4. [Optional] Add any files, folders you'd like Khoj to be aware of on the *Settings* page and Click *Save*. These files and folders will be automatically kept in sync for you # Main App You can also install the Khoj application on your desktop as a progressive web app. 1. Open the [Khoj Web App](https://app.khoj.dev) in Chrome. 2. Click on the install button in the address bar to install the app. You must be logged into your Chrome browser for this to work. Alternatively, you can also install using this route: 1. Open the three-dot menu in the top right corner of the browser. 2. Go to 'Cast, Save, and Share' option. 3. Click on the "Open in Khoj" option. --- ### Documentation Docs Clients Emacs (`documentation/docs/clients/emacs.md`) --- sidebar_position: 2 --- # Emacs Melpa Stable Badge Melpa Badge Build Badge Test Badge

> Query your Second Brain from Emacs ## Features - **Chat** - **Faster answers**: Find answers quickly, from your private notes or the public internet - **Assisted creativity**: Smoothly weave across retrieving answers and generating content - **Iterative discovery**: Iteratively explore and re-discover your notes - **Search** - **Natural**: Advanced natural language understanding using Transformer based ML Models - **Incremental**: Incremental search for a fast, search-as-you-type experience ## Interface | Search | Chat | |:------:|:----:| | | | ## Setup :::info[Self Hosting] If you are self-hosting the Khoj server modify the install steps below: - Set `khoj-server-url` to your Khoj server URL. By default, use `http://127.0.0.1:42110`. - Do not set `khoj-api-key` if your Khoj server runs in anonymous mode. For example, `khoj --anonymous-mode` ::: 1. Generate an API key on the [Khoj Web App](https://app.khoj.dev/settings#clients) 2. Add below snippet to your Emacs config file, usually at `~/.emacs.d/init.el` #### **Direct Install** *Khoj will index your org-agenda files, by default* ```elisp ;; Install Khoj.el M-x package-install khoj ; Set your Khoj API key (setq khoj-api-key "YOUR_KHOJ_CLOUD_API_KEY") (setq khoj-server-url "https://app.khoj.dev") ``` #### **Minimal Install** *Khoj will index your org-agenda files, by default* ```elisp ;; Install Khoj client from MELPA Stable (use-package khoj :ensure t :pin melpa-stable :bind ("C-c s" . 'khoj) :config (setq khoj-api-key "YOUR_KHOJ_CLOUD_API_KEY" khoj-server-url "https://app.khoj.dev")) ``` #### **Standard Install** *Configures the specified org files, directories to be indexed by Khoj* ```elisp ;; Install Khoj client from MELPA Stable (use-package khoj :ensure t :pin melpa-stable :bind ("C-c s" . 'khoj) :config (setq khoj-api-key "YOUR_KHOJ_CLOUD_API_KEY" khoj-server-url "https://app.khoj.dev" khoj-index-directories '("~/docs/org-roam" "~/docs/notes") khoj-index-files '("~/docs/todo.org" "~/docs/work.org"))) ``` #### **Straight.el** *Configures the specified org files, directories to be indexed by Khoj* ```elisp ;; Install Khoj client using Straight.el (use-package khoj :after org :straight (khoj :type git :host github :repo "khoj-ai/khoj" :files (:defaults "src/interface/emacs/khoj.el")) :bind ("C-c s" . 'khoj) :config (setq khoj-api-key "YOUR_KHOJ_CLOUD_API_KEY" khoj-server-url "https://app.khoj.dev" khoj-org-directories '("~/docs/org-roam" "~/docs/notes") khoj-org-files '("~/docs/todo.org" "~/docs/work.org"))) ``` ## Use ### Search See [Khoj Search](/features/search) for details 1. Hit `C-c s s` (or `M-x khoj RET s`) to open khoj search 2. Enter your query in natural language
E.g. *"What is the meaning of life?"*, *"My life goals for 2023"* ### Chat See [Khoj Chat](/features/chat) for details 1. Hit `C-c s c` (or `M-x khoj RET c`) to open khoj chat 2. Ask questions in a natural, conversational style
E.g. *"When did I file my taxes last year?"* ### Find Similar Entries This feature finds entries similar to the one you are currently on. 1. Move cursor to the org-mode entry, markdown section or text paragraph you want to find similar entries for 2. Hit `C-c s f` (or `M-x khoj RET f`) to find similar entries ### Advanced Usage - Add [query filters](/miscellaneous/query-filters) during search to narrow down results further e.g. `What is the meaning of life? -"god" +"none" dt>"last week"` - Use `C-c C-o 2` to open the current result at cursor in its source org file - This calls `M-x org-open-at-point` on the current entry and opens the second link in the entry. - The second link is the entries [org-id](https://orgmode.org/manual/Handling-Links.html#FOOT28), if set, or the heading text. The first link is the line number of the entry in the source file. This link is less robust to file changes. - Note: If you have [speed keys](https://orgmode.org/manual/Speed-Keys.html) enabled, `o 2` will also work ### Khoj Menu Hit `C-c s` (or `M-x khoj`) to open the khoj menu above. Then: - Hit `t` until you preferred content type is selected in the khoj menu `Content Type` specifies the content to perform `Search`, `Update` or `Find Similar` actions on - Hit `n` twice and then enter number of results you want to see `Results Count` is used by the `Search` and `Find Similar` actions - Hit `-f u` to `force` update the khoj content index The `Force Update` switch is only used by the `Update` action ## Upgrade Use your Emacs package manager to upgrade `khoj.el` #### **With MELPA** 1. Run `M-x package-refresh-content` 2. Run `M-x package-reinstall khoj` #### **With Straight.el** - Run `M-x straight-pull-package khoj` --- ### Documentation Docs Clients Obsidian (`documentation/docs/clients/obsidian.md`) --- sidebar_position: 3 --- # Obsidian > Query your Second Brain from Obsidian ## Features - **Chat** - **Faster answers**: Find answers quickly, from your private notes or the public internet - **Assisted creativity**: Smoothly weave across retrieving answers and generating content - **Iterative discovery**: Iteratively explore and re-discover your notes - **Search** - **Natural**: Advanced natural language understanding using Transformer based ML Models - **Incremental**: Incremental search for a fast, search-as-you-type experience - **Similar** - **Discover**: Find similar notes to the current one ## Setup :::info[Self Hosting] If you are self-hosting the Khoj server, update the Khoj Obsidian plugin settings step below: - Set the `Khoj URL` field to your Khoj server URL. By default, use `http://127.0.0.1:42110`. - Do not set the `Khoj API Key` field if your Khoj server runs in anonymous mode. For example, `khoj --anonymous-mode` ::: 1. Open [Khoj](https://obsidian.md/plugins?id=khoj) from the *Community plugins* tab in Obsidian settings panel 2. Click *Install*, then *Enable* on the Khoj plugin page in Obsidian 3. Generate an API key on the [Khoj Web App](https://app.khoj.dev/settings#clients) 4. Set your Khoj API Key in the Khoj plugin settings on Obsidian 5. (Optional) Click `Force Sync` in the Khoj plugin settings on Obsidian to immediately sync your Obsidian vault.
By default, your Obsidian vault is automatically synced periodically. See the official [Obsidian Plugin Docs](https://help.obsidian.md/Extending+Obsidian/Community+plugins) for more details on installing Obsidian plugins. ## Use ### Chat Click the *Khoj chat* icon πŸ’¬ on the [Ribbon](https://help.obsidian.md/User+interface/Workspace/Ribbon) or run *Khoj: Chat* from the [Command Palette](https://help.obsidian.md/Plugins/Command+palette) and ask questions in a natural, conversational style.
E.g. *"When did I file my taxes last year?"* See [Khoj Chat](/features/chat) for more details ### Find Similar Notes To see other notes similar to the current one, run *Khoj: Find Similar Notes* from the [Command Palette](https://help.obsidian.md/Plugins/Command+palette) ### Search Run *Khoj: Search* from the [Command Palette](https://help.obsidian.md/Plugins/Command+palette) See [Khoj Search](/features/search) for more details. Use [query filters](/miscellaneous/query-filters) to limit entries to search [search_demo](https://user-images.githubusercontent.com/6413477/218801155-cd67e8b4-a770-404a-8179-d6b61caa0f93.mp4 ':include :type=mp4') ## Upgrade 1. Open *Community plugins* tab in Obsidian settings 2. Click the *Check for updates* button 3. Click the *Update* button next to Khoj, if available ## Troubleshooting - Open the Khoj plugin settings pane, to configure Khoj - Toggle Enable/Disable Khoj, if setting changes have not applied - Click *Update* button to force index to refresh, if results are failing or stale --- ### Documentation Docs Clients Web (`documentation/docs/clients/web.md`) --- sidebar_position: 4 --- # Web > Query your Second Brain from your Web Browser Without any desktop clients, you can start chatting with Khoj on the web. Bear in mind you do need one of the desktop clients in order to share and sync your data with Khoj. Go see it here --> [Khoj Cloud](https://app.khoj.dev). ## Features - **Chat** - **Faster answers**: Find answers quickly, from your private notes or the public internet - **Assisted creativity**: Smoothly weave across retrieving answers and generating content - **Iterative discovery**: Iteratively explore and re-discover your notes - **Search** - **Natural**: Advanced natural language understanding using Transformer based ML Models - **Incremental**: Incremental search for a fast, search-as-you-type experience ## Setup No setup required. The Khoj web app is the default Khoj client. You can access it from any web browser. Try it on [Khoj Cloud](https://app.khoj.dev) ## Upload Documents You can upload documents to Khoj from the web interface, one at a time. This is useful for uploading documents from your phone or tablet. To upload a document: 1. You can drag and drop the document into the chat window. 2. Or click the paperclip icon in the chat window and select the document from your file system. ### Install on Phone You can optionally install Khoj as a [Progressive Web App (PWA)](https://web.dev/learn/pwa/installation). This makes it quick and easy to access Khoj on your phone. 1. Login to [Khoj Cloud](https://app.khoj.dev) or your self-hosted Khoj server from the web browser (prefer Chrome/Edge) on your phone 2. Open the 3 dot menu on the browser and click the "Add to Home screen" option 3. Click "Install" on the next screen to add the Khoj icon to your phone Home screen **Process via Screenshots** | Step 1 | Step 2 | Step 3| |:---:|:---:|:---:| | | | | --- ### Documentation Docs Clients Whatsapp (`documentation/docs/clients/whatsapp.md`) --- sidebar_position: 5 --- # WhatsApp > Query your Second Brain from WhatsApp Text [+1 (848) 800 4242](https://wa.me/18488004242) or scan the QR code below on your phone to chat with Khoj on WhatsApp. Without any desktop clients, you can start chatting with Khoj on WhatsApp. Bear in mind you do need one of the desktop clients in order to share and sync your data with Khoj. The WhatsApp AI bot will work right away for answering generic queries and using Khoj in default mode. In order to use Khoj on WhatsApp with your own data, you need to setup a Khoj Cloud account and connect your WhatsApp account to it. This is a one time setup and you can do it from the [Khoj Cloud config page](https://app.khoj.dev/settings). If you hit usage limits for the WhatsApp bot, upgrade to [a paid plan](https://khoj.dev/#pricing) on Khoj Cloud. WhatsApp QR Code ## Features - **Slash Commands**: Use slash commands to quickly access Khoj features - `/online`: Get responses from Khoj powered by online search. - `/dream`: Generate an image in response to your prompt. - `/notes`: Explicitly force Khoj to retrieve context from your notes. Note: You'll need to connect your WhatsApp account to a Khoj Cloud account for this to work. We have more commands under development, including `/share` to uploading documents directly to your Khoj account from WhatsApp, and `/speak` in order to get a speech response from Khoj. Feel free to [raise an issue](https://github.com/khoj-ai/flint/issues) if you have any suggestions for new commands. ## Source Code You can find all of the code for the WhatsApp bot in the [flint repository](https://github.com/khoj-ai/flint). As all of our code, it is open source and you can contribute to it. --- ### Documentation Docs Contributing Development (`documentation/docs/contributing/development.mdx`) --- sidebar_position: 0 --- # Development Welcome to the development docs of Khoj! Thanks for your interest in being a contributor ❀️. Open source contributors are a cornerstone of the Khoj community. We welcome all contributions, big or small. To get started with contributing, check out the official GitHub docs on [contributing to an open-source project](https://docs.github.com/en/get-started/exploring-projects-on-github/contributing-to-a-project). Join the [Discord](https://discord.gg/WaxF3SkFPU) server and click the βœ… for the question "Are you interested in becoming a contributor?" in the `#welcome-and-rules` channel. This will give you access to the `#contributors` channel where you can ask questions and get help from other contributors. If you're looking for a place to get started, check out the list of [Github Issues](https://github.com/khoj-ai/khoj/issues) with the tag `good first issue` to find issues that are good for first-time contributors. ## Local Server Installation ### Using Pip #### 1. Khoj Installation ```mdx-code-block import Tabs from '@theme/Tabs'; import TabItem from '@theme/TabItem'; ``` ```shell # Get Khoj Code git clone https://github.com/khoj-ai/khoj && cd khoj # Create, Activate Virtual Environment python3 -m venv .venv && source .venv/bin/activate # For MacOS or zsh users run this uv sync --all-extras ``` ```shell # Get Khoj Code git clone https://github.com/khoj-ai/khoj && cd khoj # Create, Activate Virtual Environment python3 -m venv .venv && .venv\Scripts\activate # Install Khoj for Development uv sync --all-extras ``` ```shell # Get Khoj Code git clone https://github.com/khoj-ai/khoj && cd khoj # Create, Activate Virtual Environment python3 -m venv .venv && source .venv/bin/activate # Install Khoj for Development uv sync --all-extras ``` #### 2. Postgres Installation & Setup Khoj uses the `pgvector` package to store embeddings of your index in a Postgres database. To use this, you need to have Postgres installed. Install [Postgres.app](https://postgresapp.com/). This comes pre-installed with `pgvector` and relevant dependencies. 1. Use the [recommended installer](https://www.postgresql.org/download/windows/). 2. Follow instructions to [Install PgVector](https://github.com/pgvector/pgvector#windows) in case you need to manually install it. Windows support is experimental for pgvector currently, so we recommend using Docker. Refer to Windows Installation Notes below if there are errors. From [official instructions](https://wiki.postgresql.org/wiki/Apt) 1. Follow instructions to [Install Postgres](https://www.postgresql.org/download/) 2. Follow instructions to [Install PgVector](https://github.com/pgvector/pgvector#installation) in case you need to manually install it. ##### Create the Khoj database Make sure to update your environment variables to match your Postgres configuration if you're using a different name. The default values should work for most people. When prompted for a password, you can use the default password `postgres`, or configure it to your preference. Make sure to set the environment variable `POSTGRES_PASSWORD` to the same value as the password you set here. ```shell createdb khoj -U postgres --password ``` ```shell createdb -U postgres khoj --password ``` ```shell sudo -u postgres createdb khoj --password ``` #### 3. Build the front-end assets ```shell cd src/interface/web/ bun install bun export ``` You can optionally use `bun dev` to start a development server for the front-end which will be available at http://localhost:3000. This is especially useful if you're making changes to the front-end code, but not necessary for running Khoj. Note that streaming does not work on the dev server due to how it is handled with SSR in Next.js. Always run `bun export` to test your front-end changes on http://localhost:42110 before creating a PR. #### 4. Run 1. Start Khoj ```bash khoj -vv ``` 2. Configure Khoj - **Via the Desktop application**: Add files, directories to index using the settings page of your desktop application. Click "Save" to immediately trigger indexing. Note: Wait after configuration for khoj to Load ML model, generate embeddings and expose API to query notes, images, documents etc specified in config YAML #### Windows Installation Notes 1. Command `khoj` Not Recognized - Try reactivating the virtual environment and rerunning the `khoj` command. - If it still doesn't work repeat the installation process. 2. Python Package Missing - Use `uv add xxx` and try running the `khoj` command. 3. Command `createdb` Not Recognized - make sure path to postgres binaries is included in environment variables. It usually looks something like ``` C:\Program Files\PostgreSQL\16\bin ``` 4. Connection Refused on port xxxx - Locate the `pg_hba.conf` file in the location where postgres was installed. - Edit the file to have **trust** as the method for user postgres, local, and host connections. - Below is an example: ``` host all postgres 127.0.0.1/32 trust # "local" is for Unix domain socket connections only local all all trust # IPv4 local connections: host all all 127.0.0.1/32 trust # IPv6 local connections: host all all ::1/128 trust ``` 4. Errors with installing pgvector - Reinstall Visual Studio 2022 Build Tools with: 1. desktop development with c++ selected in workloads 2. MSVC (C++ Build Tools), Windows 10/11 SDK, and C++/CLI support for build tools selected in individual components. - Open the x64 Native Tools Command Prompt as an Administrator - Follow the pgvector windows installation [instructions](https://github.com/pgvector/pgvector?tab=readme-ov-file#windows) in this command prompt. ### Using Docker Make sure you install the latest version of [Docker](https://docs.docker.com/get-docker/) and [Docker Compose](https://docs.docker.com/compose/install/). #### 1. Clone ```shell git clone https://github.com/khoj-ai/khoj && cd khoj ``` #### 2. Configure 1. Update [docker-compose.yml](https://github.com/khoj-ai/khoj/blob/master/docker-compose.yml) to use relevant environment variables. 2. Comment out the `image` line and uncomment the `build` line in the `server` service #### 3. Run This will start the Khoj server, and the database. ```shell docker-compose up -d ``` #### 4. Upgrade If you've made changes to the codebase, you'll need to rebuild the Docker image before running the container again. ```shell docker-compose build --no-cache ``` ## Update clients In whichever clients you're using for testing, you'll need to update the server URL to point to your local server. By default, the local server URL should be `http://127.0.0.1:42110`. ## Validate ### Before Making Changes 1. Install Khoj. Setup Git Hooks for Validation ```shell ./scripts/dev_setup.sh ``` - Installs the khoj server and web app by default. Use the `--full` flag to install the Khoj Desktop and Obsidian client apps as well. - Git hooks format code, check types automatically on every commit and push respectively. - Note 1: If [pre-commit](https://pre-commit.com/#intro) didn't already get installed, [install it](https://pre-commit.com/#install) via `pip install pre-commit` - Note 2: To run the pre-commit changes manually, use `pre-commit run --hook-stage manual --all` before creating PR ### Before Creating PR :::tip[Note] You should be in an active virtual environment for Khoj in order to run the unit tests and linter. The `dev_setup.sh` script will automatically create and activate it for you. ::: 1. Ensure that you have a [Github Issue](https://github.com/khoj-ai/khoj/issues) that can be linked to the PR. If not, create one. Make sure you've tagged one of the maintainers to the issue. This will ensure that the maintainers are notified of the PR and can review it. It's best to discuss the code design on an existing issue or Discord thread before creating a PR. This helps get your PR merged faster. 1. Run unit tests. ```shell pytest ``` 2. Run the linter. ```shell mypy ``` 4. Think about how to add unit tests to verify the functionality you're adding in the PR. If you're not sure how to do this, ask for help in the Github issue or on Discord's `#contributors` channel. ### After Creating PR 1. Automated [validation workflows](https://github.com/khoj-ai/khoj/tree/master/.github/workflows) should run for every PR. Tag one of the maintainers in the PR to trigger it. ## Obsidian Plugin Development ### Plugin development setup The core code for the Obsidian plugin is under `src/interface/obsidian`. The file `main.ts` is a good place to start. 1. In your CLI, go to the directory `src/interface/obsidian` in the Khoj repository. 2. Run `yarn install` to install the dependencies. 3. Run `yarn dev` to start the development server. This will continually rebuild the plugin as you make changes to the code. - Your code changes will be outputted to a file called `main.js` in the `obsidian` directory. ### Loading your development plugin in Obsidian 1. Make sure you have the Khoj plugin installed in Obsidian. [See the plugin page](https://publish.obsidian.md/hub/02+-+Community+Expansions/02.05+All+Community+Expansions/Plugins/khoj). 1. Open Obsidian and go to your settings (gear icon in the bottom left corner) 2. Click on 'Community Plugins' in the left panel 3. Next to the 'Installed Plugins' heading, click on the folder icon to open the folder with the plugin's source code. 4. Open the `khoj` folder in the file explorer that opens. You'll see a file called `main.js` in this folder. To test your changes, replace this file with the `main.js` file that was generated by the development server in the previous section. ## Create Khoj Release (Only for Maintainers) Follow the steps below to [release](https://github.com/khoj-ai/khoj/releases/) Khoj. This will create a stable release of Khoj on [Pypi](https://pypi.org/project/khoj/), [Melpa](https://stable.melpa.org/#%252Fkhoj) and [Obsidian](https://obsidian.md/plugins?id%253Dkhoj). It will also create desktop apps of Khoj and attach them to the latest release. 1. Create and tag release commit by running the bump_version script. The release commit sets version number in required metadata files. ```shell ./scripts/bump_version.sh -c "" ``` 2. Push commit and then the tag to trigger the release workflow to create Release with auto generated release notes. ```shell git push origin master # push release commit to khoj repository git push origin # push release tag to khoj repository ``` 3. [Optional] Update the Release Notes to highlight new features, fixes and updates ## Architecture ## Visualize Codebase *[Interactive Visualization](https://mango-dune-07a8b7110.1.azurestaticapps.net/?repo=khoj-ai%2Fkhoj)* ## Visualize Khoj Obsidian Plugin Codebase --- ### Documentation Docs Data Sources Github Integration (`documentation/docs/data-sources/github_integration.md`) # Github integration :::warning[Unmaintained] The Github integration is not maintained. We are considering deprecating it. It doesn't seem used by many folks and its cumbersome for us to maintain. ::: The Github integration allows you to index as many repositories as you want. It's currently default configured to index all Markdown/Org/Text files in each repository. For large repositories, this takes a fairly long time, but it works well for smaller projects. # Configure your settings 1. Go to [https://app.khoj.dev/settings](https://app.khoj.dev/settings) and enter in settings for the data sources you want to index. You'll have to specify the file paths. ## Use the Github plugin 1. Generate a [classic PAT (personal access token)](https://docs.github.com/en/authentication/keeping-your-account-and-data-secure/managing-your-personal-access-tokens) from [Github](https://github.com/settings/tokens) with `repo` and `admin:org` scopes at least. 2. Navigate to [https://app.khoj.dev/settings#github](https://app.khoj.dev/settings/content/github) to configure your Github settings. Enter in your PAT, along with details for each repository you want to index. 3. Click `Save`. Go back to the settings page and click `Configure`. 4. Go to [https://app.khoj.dev/](https://app.khoj.dev/) and start searching! --- ### Documentation Docs Data Sources Notion Integration (`documentation/docs/data-sources/notion_integration.md`) # Notion Integration The Notion integration allows you to search/chat with your Notion workspaces. [Notion](https://notion.so/) is a platform people use for taking notes, especially for collaboration. Go to https://app.khoj.dev/settings to connect your Notion workspace(s) to Khoj. ## Self-Hosted Setup 1. Go to https://www.notion.so/my-integrations and create a new integration called Khoj to get an API key. 3. Share all the workspaces that you want to integrate with the Khoj integration you just made in the previous step 4. In the first step, you generated an API key. Use the newly generated API Key in your Khoj settings, by default at [http://localhost:42110/settings#notion](http://localhost:42110/settings#notion). Click `Save`. 5. Click `Configure` in http://localhost:42110/settings to index your Notion workspace(s). That's it! You should be ready to start searching and chatting. Make sure you've configured your [chat settings](/get-started/setup#use-khoj). --- ### Documentation Docs Data Sources Share Your Data (`documentation/docs/data-sources/share_your_data.md`) --- sidebar_position: 0 keywords: ["upload data", "upload files", "share data", "share files", "pdf ai", "ai for pdf", "ai for documents", "ai for files", "local ai pdf", "local ai documents", "local ai files"] --- # Upload your data There are several ways you can get started with sharing your data with the Khoj AI. - Drag and drop your documents via [the web UI](/clients/web/#upload-documents). This is best if you have a one-off document you need to interact with. - You can use the [search page](https://app.khoj.dev/search) to upload documents that become indexed & searchable by your second brain. Click on "Add Documents" to get started. - Use the desktop app to [upload and sync your documents](/clients/desktop). This is best if you have a lot of documents on your computer or you need the docs to stay in sync. - Setup the sync options for either [Obsidian](/clients/obsidian) or [Emacs](/clients/emacs) to automatically sync your documents with Khoj. This is best if you are already using these tools and want to leverage Khoj's AI capabilities. - Configure your [Notion](/data-sources/notion_integration) or [Github](/data-sources/github_integration) to sync with Khoj. By providing your credentials, you can keep the data synced in the background. --- ### Documentation Docs Features Agents (`documentation/docs/features/agents.md`) --- sidebar_position: 4 --- # Agents You can use agents to setup custom system prompts with Khoj. The server host can setup their own agents, which are accessible to all users. You can see ours at https://app.khoj.dev/agents. ## Creating an Agent (Self-Hosted) Go to `server/admin/database/agent` on your server and click `Add Agent` to create a new one. You have to set it to `public` in order for it to be accessible to all the users on your server. To limit access to a specific user, do not set the `public` flag and add the user in the `Creator` field. Set your custom prompt in the `personality` field. --- ### Documentation Docs Features All Features (`documentation/docs/features/all-features.md`) --- sidebar_position: 1 --- # Overview Khoj supports a variety of features, including search and chat with a wide range of data sources and interfaces. #### [Search](/features/search) - **Local**: Your personal data stays local. All search and indexing is done on your machine when you [self-host](/get-started/setup) - **Incremental**: Incremental search for a fast, search-as-you-type experience #### [Chat](/features/chat) - **Faster answers**: Find answers faster, smoother than search. No need to manually scan through your notes to find answers. - **Iterative discovery**: Iteratively explore and (re-)discover your notes - **Assisted creativity**: Smoothly weave across answers retrieval and content generation - **Works online or offline**: Chat using online or offline AI chat models #### General - **Cloud or Self-Host**: Use [cloud](https://app.khoj.dev) to use Khoj anytime from anywhere or [self-host](/get-started/setup) for privacy - **Natural**: Advanced natural language understanding using Transformer based ML Models - **Pluggable**: Modular architecture makes it easy to plug in new data sources, frontends and ML models - **Multiple Sources**: Index your Org-mode, Markdown, PDF, plaintext files, Github repos and Notion pages - **Multiple Interfaces**: Interact from your Web Browser, Emacs, Obsidian, Desktop app or even Whatsapp ### Supported Interfaces Khoj is available as a [Desktop app](/clients/desktop), [Emacs package](/clients/emacs), [Obsidian plugin](/clients/obsidian), [Web app](/clients/web) and a [Whatsapp AI](https://khoj.dev/whatsapp). ### Supported Data Sources Khoj can understand your word, PDF, org-mode, markdown, plaintext files, and [Notion pages](/data-sources/notion_integration). --- ### Documentation Docs Features Automations (`documentation/docs/features/automations.md`) # Automations [Automations](https://app.khoj.dev/automations) are a powerful feature within Khoj to schedule repeated jobs for Khoj to do on your behalf. Automations allow you run your query automatically at a scheduled time and frequency. Khoj will send its research to you as an email. You can ask Khoj to create custom newsletters, summarize news, notify you of world events etc. Khoj uses your local time zone for scheduling. You can configure, share and delete your automations from the [automations page](https://app.khoj.dev/automations). :::info[Self Hosting] Self hosting users need to setup Resend and [Authentication](/advanced/authentication) setup to have Khoj send automation emails. ::: | [Create Automation](https://app.khoj.dev/automations?subject=Weekly%20Newsletter&query=Compile%20a%20message%20including:%201.%20A%20recap%20of%20news%20from%20last%20week%202.%20An%20at-home%20workout%20I%20can%20do%20before%20work%203.%20A%20quote%20to%20inspire%20me%20for%20the%20week%20ahead&crontime=00%209%20*%20*%201) | Receive Automation Email | |--------|-------| | create automation | create automation | --- ### Documentation Docs Features Chat (`documentation/docs/features/chat.md`) --- sidebar_position: 2 --- # Chat You can configure Khoj to chat with you about anything. When relevant, it'll use any notes or documents you shared with it to respond. It acts as an excellent research assistant, search engine, or personal tutor. Chat on Web ### Overview - Creates a personal assistant for you to inquire and engage with your notes or online information as needed - You can choose to use Online or Offline Chat depending on your requirements - Supports multi-turn conversations with the relevant notes for context - Shows reference notes used to generate a response ### Setup (Self-Hosting) See [the setup guide](/get-started/setup.mdx) to configure your chat models. ### Use 1. Open Khoj Chat - **On Web**: Open [/chat](https://app.khoj.dev/chat) in your web browser - **On Obsidian**: Search for *Khoj: Chat* in the [Command Palette](https://help.obsidian.md/Plugins/Command+palette) - **On Emacs**: Run `M-x khoj ` 2. Enter your queries to chat with Khoj. Use [slash commands](#commands) and [query filters](/miscellaneous/query-filters) to change what Khoj uses to respond #### Details 1. Your query is used to retrieve the most relevant notes, if any, using Khoj search using RAG. 2. These notes, the last few messages and associated metadata is passed to the enabled chat model along with your query to generate a response #### Conversation File Filters You can use conversation file filters to limit the notes used in the chat response. To do so, use the left panel in the web UI. Alternatively, you can also use [query filters](/miscellaneous/query-filters) to limit the notes used in the chat response. Conversation File Filter #### Commands Slash commands allows you to change what Khoj uses to respond to your query - **/notes**: Limit chat to only respond using your notes, not just Khoj's general world knowledge as reference - **/general**: Limit chat to only respond using Khoj's general world knowledge, not using your notes as reference - **/default**: Allow chat to respond using your notes or it's general knowledge as reference. It's the default behavior when no slash command is used - **/online**: Use online information and incorporate it in the prompt to the LLM to send you a response. - **/image**: Generate an image in response to your query. - **/help**: Use /help to get all available commands and general information about Khoj - **/diagram**: Generate a diagram in response to your query. This is built on [Excalidraw](https://excalidraw.com/). - **/code**: Generate and run very simple Python code snippets. Refer to [Code Execution](code_execution) for details. - **/research**: Go deeper in a topic for more accurate, in-depth responses. --- ### Documentation Docs Features Code Execution (`documentation/docs/features/code_execution.md`) --- --- # Code Execution Khoj can generate and run simple Python code as well. This is useful if you want to have Khoj do some data analysis, generate plots and reports. LLMs by default aren't skilled at complex quantitative tasks. Code generation & execution can come in handy for such tasks. Khoj automatically infers when to use the code tool. You can also tell it explicitly to use the code tool or use the `/code` [slash command](https://docs.khoj.dev/features/chat/#commands) in your chat. ## Setup (Self-Hosting) ### Terrarium Sandbox Use [Cohere's Terrarium](https://github.com/cohere-ai/cohere-terrarium) to host the code sandbox locally on your machine for free. To run with Docker, use our [docker-compose.yml](https://github.com/khoj-ai/khoj/blob/master/docker-compose.yml) to automatically setup the Terrarium code sandbox, or start it manually like this: ```bash docker pull ghcr.io/khoj-ai/terrarium:latest docker run -d -p 8080:8080 ghcr.io/khoj-ai/terrarium:latest ``` To run from source, check [these instructions](https://github.com/khoj-ai/cohere-terrarium?tab=readme-ov-file#development). #### Verify Verify that it's running, by evaluating a simple Python expression: ```bash curl -X POST -H "Content-Type: application/json" \ --url http://localhost:8080 \ --data-raw '{"code": "1 + 1"}' \ --no-buffer ``` ### E2B Sandbox [E2B](https://e2b.dev/) allows Khoj to run code on a remote but versatile sandbox with support for more python libraries. This is [not free](https://e2b.dev/pricing). To have Khoj use E2B as the code sandbox: 1. Generate an API key on [their dashboard](https://e2b.dev/dashboard). 2. Set the `E2B_API_KEY` environment variable to it on the machine running your Khoj server. - When using our [docker-compose.yml](https://github.com/khoj-ai/khoj/blob/master/docker-compose.yml), uncomment and set the `E2B_API_KEY` env var in the `docker-compose.yml` file. 3. Now restart your Khoj server to switch to using the E2B code sandbox. --- ### Documentation Docs Features Image Generation (`documentation/docs/features/image_generation.md`) # Image Generation You can use Khoj to generate images from text prompts. You can get deeper into the details of our image generation flow in this blog post: https://blog.khoj.dev/posts/how-khoj-generates-images/. To generate images, you just need to provide a prompt to Khoj in which the image generation is in the instructions. Khoj will automatically detect the image generation intent, augment your generation prompt, and then create the image. Here are some examples: | Prompt | Image | | --- | --- | | Paint a picture of the plants I got last month, pixar-animation | | | Create a picture of my dream house, based on my interests | | ## Setup (Self-Hosting) You have a couple of image generation options. ### Image Generation Models We support most state of the art image generation models, including Ideogram, Flux, and Stable Diffusion. These will run using [Replicate](https://replicate.com). Here's how to set them up: 1. Get a Replicate API key [here](https://replicate.com/account/api-tokens). 2. Create a new [Text to Image Model](http://localhost:42110/server/admin/database/texttoimagemodelconfig/). Set the `type` to `Replicate`. Use any of the model names you see [on this list](https://replicate.com/pricing#image-models). We recommend the `model name` `black-forest-labs/flux-1.1-pro` from [Replicate](https://replicate.com/black-forest-labs/flux-1.1-pro). ### OpenAI 1. Get [an OpenAI API key](https://platform.openai.com/settings/organization/api-keys). 2. Setup your OpenAI API key, if you haven't already. See instructions [here](/get-started/setup#add-chat-models) 3. Create a text to image config at http://localhost:42110/server/admin/database/texttoimagemodelconfig/. Use `model name` `dall-e-3` to use openai for image generation. Make sure to set the `Ai model api` field to the OpenAI AI model api you setup in step 2. --- ### Documentation Docs Features Keyboard Shortcuts (`documentation/docs/features/keyboard_shortcuts.md`) # Keyboard Shortcuts Oftentimes, having to leave your keyboard to move the mouse can break your flow. We want to make it as easy as possible for AI to flow with you as you are, so we've added some keyboard shortcuts to facilitate that. ## Obsidian ### Up/Down Arrow Keys (Chat Input) - **Up Arrow Key**: Move up in the list of most recent messages in your chat window. - **Down Arrow Key**: Move down in the list of most recent messages in your chat window. You can watch the demo to see how it works on [this sample conversation](http://app.khoj.dev/share/chat/in-particular-assess-the-prospect-for-brazil-/). Up/Down Arrow Keys ### Enter (Chat Input) Press 'Enter' to send the message you've typed in the chat window. --- ### Documentation Docs Features Khoj Mini (`documentation/docs/features/khoj_mini.md`) # Desktop Quick Chat (Khoj Mini) Once you have the Khoj [desktop application](https://khoj.dev/downloads) installed, you can use the desktop shortcut to quickly pull up a mini chat module for quicker answers. See the desktop setup instructions [in the docs](/clients/desktop.md) for more information. To use it, you just have to copy the text you want to inject into your query, and then run `Ctrl + Shift + K` (or `Cmd + Shift + K` on Mac) to open the mini chat module. The text you copied will be automatically pasted into the chat module, and you can then hit enter to get the answer. You can edit the text before hitting enter if you want to refine your query. The desktop shortcut is a great way to quickly get answers to your questions without having to switch between windows or tabs. It's especially useful when you're working on a project and need to quickly look up something without losing your focus. --- ### Documentation Docs Features Online Search (`documentation/docs/features/online_search.md`) # Online Search Khoj will research on the internet to ground its responses, when it determines that it would need fresh information outside its existing knowledge to answer the query. It will always show any online references it used to respond to your requests. By default, Khoj will try to infer which information sources, it needs to read to answer your question. This can include reading your documents or researching information online. You can also explicitly trigger an online search by adding the `/online` prefix to your chat query. Example queries that should trigger an online search: - What's the latest news about the Israel-Palestine war? - Where can I find the best pizza in New York City? - /online Deadline for filing taxes 2024. - Give me a summary of this article: https://en.wikipedia.org/wiki/Haitian_Revolution Try it out yourself! https://app.khoj.dev ## Self-Hosting ### Search Online search can work even with self-hosting! You have a few options: - If you're using Docker, online search should work out of the box with [searxng](https://github.com/searxng/searxng) using our standard `docker-compose.yml`. - To get production-grade, fast online search, set the `SERPER_DEV_API_KEY` environment variable to your [Serper.dev](https://serper.dev/) API key. These search results include additional context like answer box, knowledge graph etc. - To use open, self-hostable search provider, set the `FIRECRAWL_API_KEY` environment variable to your [Firecrawl](https://firecrawl.dev) API key. These search results do not scrape social media results. - To use Exa search provider, set the `EXA_API_KEY` environment variable to your [Exa](https://exa.ai) API key. ### Webpage Reading Out of the box, you **don't have to do anything to enable webpage reading**. Khoj will automatically read webpages by using the `requests` library. To get faster, more readable webpages for Khoj, you can use the following options: - For open, self-hostable webpage reader, you can use [Firecrawl](https://www.firecrawl.dev/). Create a new [Webscraper](http://localhost:42110/server/admin/database/webscraper/add/). Set your Firecrawl API key to the Api Key field, and set the type to Firecrawl. - For advanced webpage reading, you can use [Olostep](https://www.olostep.com/). This can read a wider variety of webpages. Create a new [Webscraper](http://localhost:42110/server/admin/database/webscraper/add/). Set your Olostep API key to the Api Key field, and set the type to Olostep. - For fast webpage reading, you can use [Exa](https://exa.ai). Create a new [Webscraper](http://localhost:42110/server/admin/database/webscraper/add/). Set your Exa API key to the Api Key field, and set the type to Exa. --- ### Documentation Docs Features Search (`documentation/docs/features/search.md`) --- sidebar_position: 3 --- # Search Take advantage of super fast search to find relevant notes and documents from your Second Brain. ### Use 1. Open Khoj Search - **On Web**: Open https://app.khoj.dev/ in your web browser - **On Obsidian**: Click the *Khoj search* icon πŸ”Ž on the [Ribbon](https://help.obsidian.md/User+interface/Workspace/Ribbon) or Search for *Khoj: Search* in the [Command Palette](https://help.obsidian.md/Plugins/Command+palette) - **On Emacs**: Run `M-x khoj ` 2. Query using natural language to find relevant entries from your knowledge base. Use [query filters](/miscellaneous/query-filters) to limit entries to search ### Demo ### Implementation Overview A bi-encoder models is used to create meaning vectors (aka vector embeddings) of your documents and search queries. 1. When you sync you documents with Khoj, it uses the bi-encoder model to create and store meaning vectors of (chunks of) your documents 2. When you initiate a natural language search the bi-encoder model converts your query into a meaning vector and finds the most relevant document chunks for that query by comparing their meaning vectors. 3. The slower but higher-quality cross-encoder model is than used to re-rank these documents for your given query. ### Setup (Self-Hosting) You are **not required** to configure the search model config when self-hosting. Khoj sets up decent default local search model config for general use. You may want to configure this if you need better multi-lingual search, want to experiment with different, newer models or the default models do not work for your use-case. You can use bi-encoder models downloaded locally [from Huggingface](https://huggingface.co/models?library=sentence-transformers), served via the [HuggingFace Inference API](https://endpoints.huggingface.co/), OpenAI API, Azure OpenAI API or any OpenAI Compatible API like Ollama, LiteLLM etc. Follow the steps below to configure your search model: 1. Open the [SearchModelConfig](http://localhost:42110/server/admin/database/searchmodelconfig/) page on your Khoj admin panel. 2. Hit the Plus button to add a new model config or click the id of an existing model config to edit it. 3. Set the `biencoder` field to the name of the bi-encoder model supported [locally](https://huggingface.co/models?library=sentence-transformers) or via the API you configure. 4. Set the `Embeddings inference endpoint api key` to your OpenAI API key and `Embeddings inference endpoint type` to `OpenAI` to use an OpenAI embedding model. 5. Also set the `Embeddings inference endpoint` to your Azure OpenAI or OpenAI compatible API URL to use the model via those APIs. 6. Ensure the search model config you want to use is the **only one** that has `name` field set to `default`[^1]. 7. Save the search model configs and restart your Khoj server to start using your new, updated search config. :::info You will need to re-index all your documents if you want to use a different bi-encoder model. ::: :::info You may need to tune the `Bi encoder confidence threshold` field for each bi-encoder to get appropriate number of documents for chat with your Knowledge base. Confidence here is a normalized measure of semantic distance between your query and documents. The confidence threshold limits the documents returned to chat that fall within the distance specified in this field. It can take values between 0.0 (exact overlap) and 1.0 (no meaning overlap). ::: [^1]: Khoj uses the first search model config named `default` it finds on startup as the search model config for that session --- ### Documentation Docs Features Share (`documentation/docs/features/share.md`) # Shareable Chat You can share any of your conversations by going to the three dot menu on the conversation and selecting 'Share'. This will create a **public** link that you can share with anyone. The link will open the conversation in the same state it was when you shared it, so your future messages will not be visible to the person you shared it with. This means you can easily share a conversation with someone to show them how you solved a problem, or to get help with something you're working on. --- ### Documentation Docs Features Voice Chat (`documentation/docs/features/voice-chat.md`) # Voice You can talk to Khoj using your voice. Khoj will respond to your queries using the same models as the chat feature. You can use voice chat on the web, Desktop, and Obsidian apps. Click on the little mic icon to send your voice message to Khoj. It will send back what it heard via text. You can edit the message before sending it, if required. Try it at https://app.khoj.dev/. ## Voice Response If you send a voice message, Khoj will automatically respond back with a voice message. You can also click on the speaker icon next to any message to hear it out loud. The voice response feature is available only on the web view right now. ## Setup (Self-Hosting) Voice chat will automatically be configured when you initialize the application. The default configuration will run locally. If you want to use the OpenAI whisper API for voice chat, you can set it up by following these steps: 1. Setup your OpenAI API key. See instructions [here](/get-started/setup#add-chat-models). 2. Create a new configuration at http://localhost:42110/server/admin/database/speechtotextmodeloptions/. We recommend the value `whisper-1` and model type `Openai`. If you want to use the Text to Speech feature, you can set it up by following these steps: 1. Setup your account on [ElevenLabs.io](https://elevenlabs.io/). 2. Configure your API key in your environment variables with the key `ELEVEN_LABS_API_KEY`. 3. (Optional) Create a new [Voice model option](http://localhost:42110/server/admin/database/voicemodeloption/) with a specific voice ID from whichever voice you want to use. You can explore the options [here](https://elevenlabs.io/app/voice-library). --- ### Documentation Docs Get Started Privacy Security (`documentation/docs/get-started/privacy-security.md`) --- sidebar_position: 4 slug: /privacy --- # Privacy If you're using Khoj to index you personal data, it's almost certain you'll have sensitive and private information you'd like to index. Khoj is designed to be a personal AI, so one of our cornerstone principles is to make it as privacy-friendly as possible. That's why, you can *always* choose to run Khoj on your own hardware, and never share your data outside of your device. You can generate your embeddings directly on your machine, and then use an offline chat client so that your data never leaves your machine. You'll find the instructions to [self-hosting](./setup.mdx) here. Here's what to consider if you're using Khoj, whether self-hosted or on our cloud: 1. Some of your relevant indexed data may be included as context when you chat with Khoj. This means that it may be sent to OpenAI, if you use one of the OpenAI models. 1. We collect completely anonymized usage telemetry and send it to [PostHog](https://posthog.com/). This includes data like unique chat requests, unique search requests, unique requests to index data. Usage data is collected to help us understand how people are using Khoj, and to help us prioritize features. - We do not log your IP address, nor upload any of your personal data to PostHog. - You can see our telemetry aggregation code [here](https://github.com/khoj-ai/khoj/blob/master/src/khoj/routers/helpers.py#L71) and see our telemetry server [here](https://github.com/khoj-ai/khoj/blob/master/src/telemetry/telemetry.py). - If you're self-hosting, you can opt out of telemetry by following [these instructions](/miscellaneous/telemetry). Self-hosting isn't for everyone, so we've still taken steps to make Khoj privacy-friendly, even if you choose to use our [cloud offering](https://app.khoj.dev). Here's what to consider when using Khoj Cloud: 1. Your embeddings are generated by an open source model within our own dedicated endpoint [hosted on AWS with Huggingface](https://huggingface.co/inference-endpoints/dedicated). There's zero persistent memory to the Huggingface Inference endpoints (it's stateless). 1. Your embeddings and the associated raw text are stored in a secure Postgres DB in our private AWS cloud. Your data is sharded on a unique user ID. We store the raw text in your files to improve file syncing and provide context when you chat with Khoj. 1. When you use the single-sign-on option with Google, we only receive your name, a link to your profile photo, and your email address. You can see our full privacy policy [here](https://khoj.dev/privacy-policy). :::tip[Info] Your data is yours. We do not sell your data or use it for training models. Khoj is a sustainable, open-source alternative to closed-source, commercial personal AI. We have no interest in selling your data to make a quick buck. ::: We have lots of ideas of how to make Khoj really robust as a personal AI and cloud offering, but also trust-less and privacy-centric. Please [reach out](mailto:team@khoj.dev) if this is important to you, and you'd like to help us build it. --- ### Documentation Docs Get Started Setup (`documentation/docs/get-started/setup.mdx`) --- sidebar_position: 1 --- # Self-Host Learn about how to self-host Khoj on your own machine. Benefits to self-hosting: 1. **Privacy**: Your data will never have to leave your private network. You can even use Khoj without an internet connection if deployed on your personal computer. 2. **Customization**: You can customize Khoj to your liking, from models, to host URL, to feature enablement. ```mdx-code-block import Tabs from '@theme/Tabs'; import TabItem from '@theme/TabItem'; ``` ## Setup Khoj These are the general setup instructions for self-hosted Khoj. You can install the Khoj server using either [Docker](?server=docker) or [Pip](?server=pip). :::info[First Run] Restart your Khoj server after the first run to ensure all settings are applied correctly. :::

Prerequisites

Docker

- *Option 1*: Click here to install [Docker Desktop](https://docs.docker.com/desktop/install/mac-install/). Make sure you also install the [Docker Compose](https://docs.docker.com/desktop/install/mac-install/) tool. - *Option 2*: Use [Homebrew](https://brew.sh/) to install Docker and Docker Compose. ```shell brew install --cask docker brew install docker-compose ```

Setup

1. Download the Khoj docker-compose.yml file [from Github](https://github.com/khoj-ai/khoj/blob/master/docker-compose.yml) ```shell mkdir ~/.khoj && cd ~/.khoj wget https://raw.githubusercontent.com/khoj-ai/khoj/master/docker-compose.yml ``` 2. Configure the environment variables in the `docker-compose.yml` - Set `KHOJ_ADMIN_PASSWORD`, `KHOJ_DJANGO_SECRET_KEY` (and optionally the `KHOJ_ADMIN_EMAIL`) to something secure. This allows you to customize Khoj later via the admin panel. - Set `OPENAI_API_KEY`, `ANTHROPIC_API_KEY`, or `GEMINI_API_KEY` to your API key if you want to use OpenAI, Anthropic or Gemini commercial chat models respectively. - Uncomment `OPENAI_BASE_URL` to use [Ollama](/advanced/ollama?type=first-run&server=docker#setup) running on your host machine. Or set it to the URL of your OpenAI compatible API like vLLM or [LMStudio](/advanced/lmstudio). 3. Start Khoj by running the following command in the same directory as your docker-compose.yml file. ```shell cd ~/.khoj docker-compose up ```

Prerequisites

1. Install [WSL2](https://learn.microsoft.com/en-us/windows/wsl/install) and restart your machine ```shell # Run in PowerShell wsl --install ``` 2. Install [Docker Desktop](https://docs.docker.com/desktop/install/windows-install/) with **[WSL2 backend](https://docs.docker.com/desktop/wsl/#turn-on-docker-desktop-wsl-2)** (default)

Setup

1. Download the Khoj docker-compose.yml file [from Github](https://github.com/khoj-ai/khoj/blob/master/docker-compose.yml) ```shell # Windows users should use their WSL2 terminal to run these commands mkdir ~/.khoj && cd ~/.khoj wget https://raw.githubusercontent.com/khoj-ai/khoj/master/docker-compose.yml ``` 2. Configure the environment variables in the `docker-compose.yml` - Set `KHOJ_ADMIN_PASSWORD`, `KHOJ_DJANGO_SECRET_KEY` (and optionally the `KHOJ_ADMIN_EMAIL`) to something secure. This allows you to customize Khoj later via the admin panel. - Set `OPENAI_API_KEY`, `ANTHROPIC_API_KEY`, or `GEMINI_API_KEY` to your API key if you want to use OpenAI, Anthropic or Gemini commercial chat models respectively. - Uncomment `OPENAI_BASE_URL` to use [Ollama](/advanced/ollama) running on your host machine. Or set it to the URL of your OpenAI compatible API like vLLM or [LMStudio](/advanced/lmstudio). 3. Start Khoj by running the following command in the same directory as your docker-compose.yml file. ```shell # Windows users should use their WSL2 terminal to run these commands cd ~/.khoj docker-compose up ```

Prerequisites

Install [Docker Desktop](https://docs.docker.com/desktop/install/linux/). You can also use your package manager to install Docker Engine & Docker Compose.

Setup

1. Download the Khoj docker-compose.yml file [from Github](https://github.com/khoj-ai/khoj/blob/master/docker-compose.yml) ```shell mkdir ~/.khoj && cd ~/.khoj wget https://raw.githubusercontent.com/khoj-ai/khoj/master/docker-compose.yml ``` 2. Configure the environment variables in the `docker-compose.yml` - Set `KHOJ_ADMIN_PASSWORD`, `KHOJ_DJANGO_SECRET_KEY` (and optionally the `KHOJ_ADMIN_EMAIL`) to something secure. This allows you to customize Khoj later via the admin panel. - Set `OPENAI_API_KEY`, `ANTHROPIC_API_KEY`, or `GEMINI_API_KEY` to your API key if you want to use OpenAI, Anthropic or Gemini commercial chat models respectively. - Uncomment `OPENAI_BASE_URL` to use [Ollama](/advanced/ollama) running on your host machine. Or set it to the URL of your OpenAI compatible API like vLLM or [LMStudio](/advanced/lmstudio). 3. Start Khoj by running the following command in the same directory as your docker-compose.yml file. ```shell cd ~/.khoj docker-compose up ``` :::info[Remote Access] By default Khoj is only accessible on the machine it is running. To access Khoj from a remote machine see [Remote Access Docs](/advanced/remote). ::: Your setup is complete once you see `πŸŒ– Khoj is ready to engage` in the server logs on your terminal.

1. Install Khoj Server

- Make sure [python](https://realpython.com/installing-python/) and [pip](https://pip.pypa.io/en/stable/installation/) are installed on your machine - Check [llama-cpp-python setup](https://github.com/abetlen/llama-cpp-python?tab=readme-ov-file#supported-backends) if you hit any llama-cpp issues with the installation Run the following command in your terminal to install the Khoj server. ```shell CMAKE_ARGS="-DGGML_METAL=on" python -m pip install 'khoj[local]' ``` ```shell python -m pip install 'khoj[local]' ``` Run the following command in PowerShell on Windows ```shell # Install Khoj py -m pip install 'khoj[local]' ``` ```shell # 1. To use NVIDIA (CUDA) GPU $env:CMAKE_ARGS = "-DGGML_CUDA=on" # 2. Install Khoj py -m pip install 'khoj[local]' ``` ```shell # 1. To use AMD (ROCm) GPU $env:CMAKE_ARGS = "-DGGML_HIPBLAS=on" # 2. Install Khoj py -m pip install 'khoj[local]' ``` ```shell # 1. To use VULCAN GPU $env:CMAKE_ARGS = "-DGGML_VULKAN=on" # 2. Install Khoj py -m pip install 'khoj[local]' ``` ```shell python -m pip install 'khoj[local]' ``` ```shell CMAKE_ARGS="-DGGML_CUDA=on" FORCE_CMAKE=1 python -m pip install 'khoj[local]' ``` ```shell CMAKE_ARGS="-DGGML_HIPBLAS=on" FORCE_CMAKE=1 python -m pip install 'khoj[local]' ``` ```shell CMAKE_ARGS="-DGGML_VULKAN=on" FORCE_CMAKE=1 python -m pip install 'khoj[local]' ```

2. Start Khoj Server

Run the following command from your terminal to start the Khoj service. ```shell USE_EMBEDDED_DB="true" khoj --anonymous-mode ``` `--anonymous-mode` allows access to Khoj without requiring login. This is usually fine for local only, single user setups. If you need authentication follow the [authentication setup docs](/advanced/authentication).

First Run

On the first run of the above command, you will be prompted to: 1. Create an admin account with an email and secure password 2. Customize the chat models to enable - Keep your [OpenAI](https://platform.openai.com/api-keys), [Anthropic](https://console.anthropic.com/account/keys), [Gemini](https://aistudio.google.com/app/apikey) API keys and [OpenAI](https://platform.openai.com/docs/models), [Anthropic](https://docs.anthropic.com/en/docs/about-claude/models#model-names), [Gemini](https://cloud.google.com/vertex-ai/generative-ai/docs/learn/models#gemini-models), [Offline](https://huggingface.co/models?pipeline_tag=text-generation&library=gguf) chat model names handy to set any of them up during first run. 3. Your setup is complete once you see `πŸŒ– Khoj is ready to engage` in the server logs on your terminal! :::tip[Auto Start] To start Khoj automatically in the background use [Task scheduler](https://www.windowscentral.com/how-create-automated-task-using-task-scheduler-windows-10) on Windows or [Cron](https://en.wikipedia.org/wiki/Cron) on Mac, Linux (e.g. with `@reboot khoj`) ::: ## Use Khoj You can now open the web app at http://localhost:42110 and start interacting!
Nothing else is necessary, but you can customize your setup further by following the steps below. ### Add Chat Models

Login to the Khoj Admin Panel

Go to http://localhost:42110/server/admin and login with the admin credentials you setup during installation. :::info[CSRF Error] Ensure you are using **localhost, not 127.0.0.1**, to access the admin panel to avoid the CSRF error. ::: :::info[CSRF Trusted Origin or Unset Cookie Error] If using a load balancer/reverse_proxy in front of your Khoj server: Set the environment variable KHOJ_ALLOWED_DOMAIN=your-internal-ip-or-domain to avoid this error. If unset, it defaults to KHOJ_DOMAIN. ::: :::info[DISALLOWED HOST or Bad Request (400) Error] You may hit this if you try access Khoj exposed on a custom domain (e.g. 192.168.12.3 or example.com) or over HTTP. Set the environment variables KHOJ_DOMAIN=your-external-ip-or-domain and KHOJ_NO_HTTPS=True if required to avoid this error. ::: :::tip[Note] Using Safari on Mac? You might not be able to login to the admin panel. Try using Chrome or Firefox instead. :::

Configure Chat Model

Setup which chat model you'd want to use. Khoj supports local and online chat models. :::info[Ollama Integration] Using Ollama? See the [Ollama Integration](/advanced/ollama) section for more custom setup instructions. ::: 1. Create a new [AI Model Api](http://localhost:42110/server/admin/database/aimodelapi/add) in the server admin settings. - Add your [OpenAI API key](https://platform.openai.com/api-keys) - Give the configuration a friendly name like `OpenAI` - (Optional) Set the API base URL. It is only relevant if you're using another OpenAI-compatible proxy server like [Ollama](/advanced/ollama) or [LMStudio](/advanced/lmstudio).
2. Create a new [chat model](http://localhost:42110/server/admin/database/chatmodel/add) - Set the `chat-model` field to an [OpenAI chat model](https://platform.openai.com/docs/models). Example: `gpt-4o`. - Make sure to set the `model-type` field to `OpenAI`. - If your model supports vision, set the `vision enabled` field to `true`. This is currently only supported for OpenAI models with vision capabilities. - The `tokenizer` and `max-prompt-size` fields are optional. Set them only if you're sure of the tokenizer or token limit for the model you're using. Contact us if you're unsure what to do here.
 1. Create a new [AI Model API](http://localhost:42110/server/admin/database/aimodelapi/add) in the server admin settings. - Add your [Anthropic API key](https://console.anthropic.com/account/keys) - Give the configuration a friendly name like `Anthropic`. Do not configure the API base url. 2. Create a new [chat model](http://localhost:42110/server/admin/database/chatmodel/add) - Set the `chat-model` field to an [Anthropic chat model](https://docs.anthropic.com/en/docs/about-claude/models#model-names). Example: `claude-3-5-sonnet-20240620`. - Set the `model-type` field to `Anthropic`. - Set the `ai model api` field to the Anthropic AI Model API you created in step 1. 1. Create a new [AI Model API](http://localhost:42110/server/admin/database/aimodelapi/add) in the server admin settings. - Add your [Gemini API key](https://aistudio.google.com/app/apikey) - Give the configuration a friendly name like `Gemini`. Do not configure the API base url. 2. Create a new [chat model](http://localhost:42110/server/admin/database/chatmodel/add) - Set the `chat-model` field to a [Google Gemini chat model](https://cloud.google.com/vertex-ai/generative-ai/docs/learn/models#gemini-models). Example: `gemini-2.5-flash`. - Set the `model-type` field to `Google`. - Set the `ai model api` field to the Gemini AI Model API you created in step 1. Offline chat stays completely private and can work without internet using any open-weights model. :::tip[System Requirements] - Minimum 8 GB RAM. Recommend **16Gb VRAM** - Minimum **5 GB of Disk** available - A Nvidia, AMD GPU or a Mac M1+ machine would significantly speed up chat responses ::: 1. Install any Openai API compatible local ai model server like [llama-cpp-server](https://github.com/ggml-org/llama.cpp/tree/master/tools/server), Ollama, vLLM etc. 2. Add an [ai model api](http://localhost:42110/server/admin/database/aimodelapi/add/) on the admin panel - Set the `api url` field to the url of your local ai model provider like `http://localhost:11434/v1/` for Ollama 3. Restart the Khoj server to load models available on your local ai model provider - If that doesn't work, you'll need to manually add available [chat model](http://localhost:42110/server/admin/database/chatmodel/add) in the admin panel. 4. Set the newly added chat model as your preferred model in your [User chat settings](http://localhost:42110/settings) 5. [Start chatting](http://localhost:42110) with your local AI! :::tip[Multiple Chat Models] Set your preferred default chat model in the `Default`, `Advanced` fields of your [ServerChatSettings](http://localhost:42110/server/admin/database/serverchatsettings/). Khoj uses these chat model for all intermediate steps like intent detection, web search etc. ::: :::info[Chat Model Fields] - The `tokenizer` and `max-prompt-size` fields are optional. Set them only if you're sure of the tokenizer or token limit for the model you're using. This improves context stuffing. Contact us if you're unsure what to do here. - Only tick the `vision enabled` field for OpenAI models with vision capabilities like gpt-4o. Vision capabilities in other chat models is not currently utilized. ::: ### Sync your Knowledge - You can chat with your notes and documents using Khoj. - Khoj can keep your files and folders synced using the Khoj [Desktop](/clients/desktop#setup), [Obsidian](/clients/obsidian#setup) or [Emacs](/clients/emacs#setup) clients. - Your [Notion workspace](/data-sources/notion_integration) can be directly synced from the web app. - You can also just drag and drop specific files you want to chat with on the [Web app](/clients/web#upload-documents). ### Setup Khoj Clients The Khoj web app is available by default to chat, search and configure Khoj.
You can also install a Khoj client to easily access it from Obsidian, Emacs, Whatsapp or your OS and keep your documents synced with Khoj. :::info[Note] Set the host URL on your clients settings page to your Khoj server URL. By default, use `http://127.0.0.1:42110` or `http://localhost:42110`. Note that `localhost` may not work in all cases. ::: - Read the Khoj Desktop app [setup docs](/clients/desktop#setup). - Read the Khoj Emacs package [setup docs](/clients/emacs#setup). - Read the Khoj Obsidian plugin [setup docs](/clients/obsidian#setup). - Read the Khoj Whatsapp app [setup docs](/clients/whatsapp). ## Upgrade ### Upgrade Server ```shell pip install --upgrade khoj ``` *Note: To upgrade to the latest pre-release version of the khoj server run below command* Run the commands below from the same directory where you have your `docker-compose.yml` file. This will fetch the latest build and upgrade your server. ```shell # Windows users should use their WSL2 terminal to run these commands cd ~/.khoj # assuming your khoj docker-compose.yml file is here docker-compose up --build ``` ### Upgrade Clients - The Desktop app automatically updates to the latest released version on restart. - You can manually download the latest version from the [Khoj Website](https://khoj.dev/downloads). - Use your Emacs Package Manager to Upgrade - See [khoj.el package setup](/clients/emacs#setup) for details - Upgrade via the Community plugins tab on the settings pane in the Obsidian app - See the [khoj plugin setup](/clients/obsidian#setup) for details ## Uninstall ### Uninstall Server ```shell # uninstall khoj server pip uninstall khoj # delete khoj postgres db dropdb khoj -U postgres ``` Run the command below from the same directory where you have your `docker-compose` file. This will remove the server containers, networks, images and volumes. ```shell docker-compose down --volumes ``` ### Uninstall Clients Uninstall the Khoj Desktop client in the standard way from your OS. Uninstall the Khoj Emacs package in the standard way from Emacs. Uninstall via the Community plugins tab on the settings pane in the Obsidian app ## Troubleshoot #### Dependency conflict when trying to install Khoj python package with pip - **Reason**: When conflicting dependency versions are required by Khoj vs other python packages installed on your system - **Fix**: Install Khoj in a python virtual environment using [venv](https://docs.python.org/3/library/venv.html) or [pipx](https://pypa.github.io/pipx) to avoid this dependency conflicts - **Process**: 1. Install [pipx](https://pypa.github.io/pipx/#install-pipx) 2. Use `pipx` to install Khoj to avoid dependency conflicts with other python packages. ```shell pipx install khoj ``` 3. Now start `khoj` using the standard steps described earlier #### Install fails while building Tokenizer dependency - **Details**: `pip install khoj` fails while building the `tokenizers` dependency. Complains about Rust. - **Fix**: Install Rust to build the tokenizers package. For example on Mac run: ```shell brew install rustup rustup-init source ~/.cargo/env ``` - **Refer**: [Issue with Fix](https://github.com/khoj-ai/khoj/issues/82#issuecomment-1241890946) for more details #### Khoj in Docker errors out with \"Killed\" in error message - **Fix**: Increase RAM available to Docker Containers in Docker Settings - **Refer**: [StackOverflow Solution](https://stackoverflow.com/a/50770267), [Configure Resources on Docker for Mac](https://docs.docker.com/desktop/mac/#resources) --- ### Documentation Docs Miscellaneous Credits (`documentation/docs/miscellaneous/credits.md`) --- sidebar_position: 4 --- # Credits Many Open Source projects are used to power Khoj. Here's a few of them: - [Llama.cpp](https://github.com/ggerganov/llama.cpp) to chat with local LLM - [SentenceTransformer](https://www.sbert.net/examples/applications/retrieve_rerank/README.html) for Text Search - [HuggingFace](https://huggingface.co/) for hosting open-source chat and search models - Charles Cave for [OrgNode Parser](http://members.optusnet.com.au/~charles57/GTD/orgnode.html) - [Org.js](https://mooz.github.io/org-js/) to render Org-mode results on the Web interface - [Markdown-it](https://github.com/markdown-it/markdown-it) to render Markdown results on the Web interface - [Katex](https://katex.org/) to render math --- ### Documentation Docs Miscellaneous Performance (`documentation/docs/miscellaneous/performance.md`) --- sidebar_position: 2 --- # Performance Here are some top-level performance metrics for Khoj. These are rough estimates and will vary based on your hardware and data. :::info These performance metrics were last evaluated in 2022. ::: ### Search performance - Semantic search using the default embeddings model is fairly fast at \<100 ms across all content types - Reranking using the cross-encoder model is slower at \<2s on 15 results. Tweak `top_k` to tradeoff speed for accuracy of results - Filters in query (e.g. by file, word or date) usually add \<20ms to query latency ### Indexing performance - Indexing is more strongly impacted by the size of the source data - Indexing 100K+ line corpus of notes takes about 10 minutes - Note: *It should only take this long on the first run* as the index is incrementally updated ### Miscellaneous - Testing done on a Mac M1 and a \>100K line corpus of notes --- ### Documentation Docs Miscellaneous Query Filters (`documentation/docs/miscellaneous/query-filters.md`) --- sidebar_position: 3 --- # Query Filters Use structured query syntax to filter entries from your knowledge based used by search results or chat responses. - **Word Filter**: Get entries that include/exclude a specified term - Entries that contain term_to_include: `+"term_to_include"` - Entries that contain term_to_exclude: `-"term_to_exclude"` - **Date Filter**: Get entries containing dates in YYYY-MM-DD format from specified date (range) - Entries from April 1st 1984: `dt:"1984-04-01"` - Entries after March 31st 1984: `dt>="1984-04-01"` - Entries before April 2nd 1984 : `dt<="1984-04-01"` - **File Filter**: Get entries from a specified file - Entries from incoming.org file: `file:"incoming.org"` - Combined Example - `what is the meaning of life? file:"1984.org" dt>="1984-01-01" dt<="1985-01-01" -"big" -"brother"` - Adds all filters to the natural language query. It should return entries - from the file *1984.org* - containing dates from the year *1984* - excluding words *"big"* and *"brother"* - that best match the natural language query *"what is the meaning of life?"* --- ### Documentation Docs Miscellaneous Telemetry (`documentation/docs/miscellaneous/telemetry.md`) --- sidebar_position: 1 --- # Telemetry We collect some high level, anonymized metadata about usage of self-hosted Khoj. This includes: - Client (Web, Emacs, Obsidian) - API usage (Search, Chat) - Configured content types (Github, Org, etc) - Request metadata (e.g., host, referrer) We don't send any personal information or any information from/about your content. We only send the above metadata. This helps us prioritize feature development and understand how people are using Khoj. Don't just take our word for it -- you can see [the code here](https://github.com/khoj-ai/khoj/tree/master/src/telemetry). ## Disable Telemetry If you're self-hosting Khoj, you can opt out of telemetry at any time by setting the `KHOJ_TELEMETRY_DISABLE` environment variable to `True` via shell or `docker-compose.yml` If you have any questions or concerns, please reach out to us on [Discord](https://discord.gg/BDgyabRM6e). --- --- METRICS --- - Files Extracted: 40 - Estimated Token Budget: ~29453 tokens - Canonical Reference: https://codewiki.google/github.com/khoj-ai/khoj