{"id":13408935,"url":"https://github.com/sedwards2009/llm-multitool","last_synced_at":"2025-10-29T15:31:40.145Z","repository":{"id":205624963,"uuid":"658117196","full_name":"sedwards2009/llm-multitool","owner":"sedwards2009","description":"Web UI for working with large language models","archived":false,"fork":false,"pushed_at":"2024-02-24T16:20:12.000Z","size":3737,"stargazers_count":18,"open_issues_count":1,"forks_count":1,"subscribers_count":3,"default_branch":"main","last_synced_at":"2024-02-24T17:29:12.005Z","etag":null,"topics":["ai","artificial-intelligence","chatgpt","chatgpt-api","go","golang","llm","ollama"],"latest_commit_sha":null,"homepage":"","language":"Go","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/sedwards2009.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null}},"created_at":"2023-06-24T20:27:14.000Z","updated_at":"2024-02-10T22:32:01.000Z","dependencies_parsed_at":null,"dependency_job_id":"abb45c30-2c5e-41ba-a6da-8903eb8ff20c","html_url":"https://github.com/sedwards2009/llm-multitool","commit_stats":null,"previous_names":["sedwards2009/llm-workbench","sedwards2009/llm-multitool"],"tags_count":3,"template":false,"template_full_name":null,"purl":"pkg:github/sedwards2009/llm-multitool","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/sedwards2009%2Fllm-multitool","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/sedwards2009%2Fllm-multitool/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/sedwards2009%2Fllm-multitool/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/sedwards2009%2Fllm-multitool/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/sedwards2009","download_url":"https://codeload.github.com/sedwards2009/llm-multitool/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/sedwards2009%2Fllm-multitool/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":281646118,"owners_count":26537234,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","status":"online","status_checked_at":"2025-10-29T02:00:06.901Z","response_time":59,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["ai","artificial-intelligence","chatgpt","chatgpt-api","go","golang","llm","ollama"],"created_at":"2024-07-30T20:00:56.720Z","updated_at":"2025-10-29T15:31:39.458Z","avatar_url":"https://github.com/sedwards2009.png","language":"Go","funding_links":[],"categories":["📚 Contents","List"],"sub_categories":["ChatGPT"],"readme":"# llm-multitool\n\n**llm-multitool** is a local web UI for working with large language models (LLM). It oriented towards *instruction tasks* and can connect to and use different servers running LLMs.\n\n\n![screenshot](screenshot.png)\n\n## Features\n\n* Aims to be easy to use\n* Supports different LLM backends/servers including locally run ones:\n  * [OpenAI's ChatGPT](https://openai.com/chatgpt)\n  * [Ollama](https://github.com/jmorganca/ollama)\n  * and most backends which support the OpenAI API such as [LocalAI](https://localai.io/) and [Oobabooga](https://github.com/oobabooga/).\n* \"Instruction\" templates to simplify certain tasks.\n* Chat support for backends which support it.\n* Multiple persistent sessions and history\n\n\n## Downloading\n\nExecutables for Windows, macOS, and Linux can be downloaded from the [Releases page](https://github.com/sedwards2009/llm-multitool/releases).\n\n## Building from source\n\nInstead of downloading a precompiled executable you can also build it from source.\n\n### Build Dependencies\n\n* This project uses [Taskfile](https://taskfile.dev/) for building. The `task` executable must be available in your path. You can install the `task` binary from https://taskfile.dev/ .\n* [Go](https://go.dev/) compiler\n* Recent [Node JS](https://nodejs.org/) version.\n\n### Building\n\nRun `task prepare` and `task build` to prepare and build llm-multitool. It will first build the web frontend and then the Go based backend. The output executable will be in the `backend/` folder and named `llm-multitool`.\n\n## Backends\n\nllm-multitool is just a UI for LLMs. It needs a LLM backend to connect to which actually runs the model. But which one should you use?\n\nShort answer:\n\n* If you don't want to run a LLM locally then you should set up OpenAI's ChatGPT.\n* If you do want to run a LLM locally then try Ollama.\n\nThe backends in more detail:\n\n* **[OpenAI's ChatGPT](https://openai.com/chatgpt)** - Support for this backend is the most complete and stable. It does require setting up billing and an API token at OpenAI to use.\n* **[Ollama](https://ollama.ai/)** - This can run LLMs locally, is easy to set up, and supports Linux, Windows, and macOS. Version v0.1.14 or later is required.\n* **[LocalAI](https://localai.io/)** - This will also let you run LLMs locally, is easy to set up and supports many different LLMs, but only runs Linux and macOS. llm-multitool supports this quite well via it's OpenAI API.\n* **[Oobabooga text-generation-ui](https://github.com/oobabooga/)** - This backend can be a challenge to install and isn't really meant for end users. It does support have many LLM types. llm-multitool support for this mostly works but is buggy.\n\nNote: It should be possible to connect llm-multitool to most things which support the OpenAI API.\n\n\n## Configuring a backend\n\nBefore running llm-multitool you first need to write a small configuration file in yaml to set up which backend(s) it should connect to and how. By default you can name this file `backend.yaml`. Its name can be specified when starting llm-multitool.\n\nThis configuration file consists of 1 or more backend configurations in a YAML list. You may have as many backends configured as you want. At start up llm-multitool will query each backend for its list of models. If a backend is not available, then it is jus skipped.\n\nBelow are sections on how to configure the different backends.\n\n### OpenAI Configuration\n\nThe following example `backend.yaml` file shows how to connect to OpenAI's ChatGPT model. You need to generate your own API token on OpenAI's website to use in the file.\n\n```yaml\n- name: OpenAI\n  api_token: \"sk-FaKeOpEnAiToKeN7Nll3FAKzZET3BlbkFJLz8Oume19ZeAjGh3rabc\"\n  models:\n  - gpt-3.5-turbo\n  - gpt-4\n```\nThe `name` field can be any name you like, but it is best to keep it short.\n\n`api_token` holds the value of the token you generated at OpenAI.\n\nIf you don't want to copy your token directly into your configuration file, you can omit the `api_token` file and replace it with `api_token_from` with a value naming a text file from which to read the read token from. The file path is relative to the `backend.yaml` file.\n\n`models` is a list of model to permit. OpenAI have many different models and varieties, but only a handful of the the LLMs are useful for use with llm-multitool.\n\n### Ollama\n\nllm-multitool can connect to a Ollama server via its own API. The configuration block is as follows:\n\n```yaml\n- name: Ollama\n  address: \"http://localhost:11434\"\n  variant: ollama\n```\n\nThe `name` field can be any name you like, but it is best to keep it short.\n\nThe `address` field is the URL of the Ollama server.\n\nThe `variant` field must have the value \"ollama\".\n\n\n### LocalAI\n\nLocalAI and OpenAI configuration is the same thing except that LocalAI needs an `address` value to be specified and it doesn't require the `token` or `model` values.\n\n```yaml\n- name: LocalAI\n  address: \"http://127.0.0.1:5001/v1\"\n```\n\nThe `address` field is the URL of the LocalAI server.\n\n### Oobabooga text-generation-ui\n\nllm-multitool can connect to a running Oobabooga text-generation-ui server via its OpenAI extension.\n\n⚠️ You need to set up and activate the `openai` extension in your Oobabooga installation. The `README.md` file under Oobabooga's `extensions/openai/` folder gives more details. Also, when starting Oobabooga you need to turn the extension on.\n\nThe configuration needed in `backend.yaml` is typically:\n\n```yaml\n- name: Ooba\n  address: \"http://127.0.0.1:5001/v1\"\n  variant: oobabooga\n```\n\nThe `name` field can be any name you like, but it is best to keep it short.\n\nThe `address` field is the URL of the OpenAI end-point running on Oobabooga.\n\nThe `variant` field must have the value \"oobabooga\".\n\n## Running\n\nIf you have built llm-multitool from source then the executable will be in `backend/llm-multitool` and you should have written a minial `backend.yaml` file. Start up `llm-multitool` with:\n\n    backend/llm-multitool -c backend.yaml\n\nThis will start the server and it will listen on address `127.0.0.1` port `5050` by default.\n\nOpen your browser on http://127.0.0.1:5050 to use the llm-multitool UI.\n\n## Command line reference\n\n    usage: llm-multitool [-h|--help] [-c|--config \"\u003cvalue\u003e\"] [-s|--storage\n    \"\u003cvalue\u003e\"] [-p|--presets \"\u003cvalue\u003e\"] [-t|--templates\n    \"\u003cvalue\u003e\"] [-a|--address \"\u003cvalue\u003e\"]\n\n    Web UI for instructing Large Language Models\n\n    Arguments:\n\n    -h  --help       Print help information\n    -c  --config     Path to the configuration file. Default: backend.yaml\n    -s  --storage    Path to the session data storage directory. Default: data\n    -p  --presets    Path to the file containing generation parameter presets.\n    Default:\n    -t  --templates  Path to the file containing templates. Default:\n    -a  --address    Address and port to server from. Default: 127.0.0.1:5050\n\n\n## Custom instruction templates\n\nllm-multitool has a small set of built in templates for instruct type tasks. You can read this yaml file up on GitHub [here](https://github.com/sedwards2009/llm-multitool/blob/main/backend/config/templates.yaml). It is possible to create your own templates file and tell llm-multitool to use it with the `-t` command line option.\n\nThe format of the templates yaml file is an array of template objects. Each template object has the following fields:\n\n* `id` - A unique string to identify the template. UUIDs work well here, but any string is accepted\n* `name` - The name of the template. This will be shown in the web UI. For example, \"Translate to French\"\n* `template_string` - The template for the prompt itself. The string `{{prompt}}` will be replaced with what ever the user enters as the prompt in the web UI.\n\nIf you write an interesting template, consider submitting it to this project for inclusion.\n\n## Custom parameter presets\n\nllm-multitool has a small set of built in parameter presets. These control the generation of responses. You can read this yaml file up on GitHub [here](https://github.com/sedwards2009/llm-multitool/blob/main/backend/config/presets.yaml). It is possible to create your own presets file and tell llm-multitool to use it with the `-p` command line option.\n\n* `id` - A unique string to identify the preset.\n* `name` - The name of the preset. This will be shown in the web UI.\n* `temperature` - a numeric value specifying the temperature value to use during generation. For example, 0.8\n* `top_p` - a numeric value specifying the Top P setting to use during generation. For example, 0.1\n\n\n## License\n\nMIT\n\n## Author\n\nSimon Edwards \u003csimon@simonzone.com\u003e\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fsedwards2009%2Fllm-multitool","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fsedwards2009%2Fllm-multitool","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fsedwards2009%2Fllm-multitool/lists"}