|
1 | 1 | # GPTScript |
2 | 2 |
|
3 | | -[](https://discord.gg/9sSf4UyAMC) |
| 3 | + |
4 | 4 |
|
5 | | -## Overview |
| 5 | +GPTScript is a framework that allows Large Language Models (LLMs) to operate and interact with various systems. These systems can range from local executables to complex applications with OpenAPI schemas, SDK libraries, or any RAG-based solutions. GPTScript is designed to easily integrate any system, whether local or remote, with your LLM using just a few lines of prompts. |
6 | 6 |
|
7 | | -GPTScript is a new scripting language to automate your interaction with a Large Language Model (LLM), namely OpenAI. The ultimate goal is to create a natural language programming experience. The syntax of GPTScript is largely natural language, making it very easy to learn and use. |
8 | | -Natural language prompts can be mixed with traditional scripts such as bash and python or even external HTTP service |
9 | | -calls. With GPTScript you can do just about anything, like [plan a vacation](./examples/travel-agent.gpt), |
10 | | -[edit a file](./examples/add-go-mod-dep.gpt), [run some SQL](./examples/sqlite-download.gpt), or [build a mongodb/flask app](./examples/hacker-news-headlines.gpt). Here are some common use cases for GPTScript: |
| 7 | +Here are some sample use cases of GPTScript: |
| 8 | +1. Chat with a local CLI - [Try it!](https://docs.gptscript.ai/examples/cli) |
| 9 | +2. Chat with an OpenAPI compliant endpoint - [Try it!](https://docs.gptscript.ai/examples/api) |
| 10 | +3. Chat with local files and directories - [Try it!](https://docs.gptscript.ai/examples/local-files) |
| 11 | +4. Run an automated workflow - [Try it!](https://docs.gptscript.ai/examples/workflow) |
11 | 12 |
|
12 | | -1. [Retrieval-Augmented Generation (RAG)](./docs/README-USECASES.md#retrieval) |
13 | | -2. [Task Automation](./docs/README-USECASES.md#task-automation) |
14 | | -3. [Agents and Assistants](./docs/README-USECASES.md#agents-and-assistants) |
15 | | -4. [Data Analysis](./docs/README-USECASES.md#data-analysis) |
16 | | -5. [Vision, Image, and Audio](./docs/README-USECASES.md#vision-image-and-audio) |
17 | | -6. [Memory Management](./docs/README-USECASES.md#memory-management) |
18 | | -7. [Chatbots](./docs/README-USECASES.md#chatbots) |
19 | 13 |
|
20 | | -| :memo: | We are currently exploring options for interacting with local models using GPTScript. | |
21 | | -| ------ | :------------------------------------------------------------------------------------ | |
22 | | - |
23 | | -The following example illustrates how GPTScript allows you to accomplish a complex task by writing instructions in English: |
24 | | - |
25 | | -```yaml |
26 | | -# example.gpt |
27 | | - |
28 | | -Tools: sys.download, sys.exec, sys.remove |
29 | | - |
30 | | -Download https://www.sqlitetutorial.net/wp-content/uploads/2018/03/chinook.zip to a |
31 | | -random file. Then expand the archive to a temporary location as there is a sqlite |
32 | | -database in it. |
33 | | - |
34 | | -First inspect the schema of the database to understand the table structure. |
35 | | - |
36 | | -Form and run a SQL query to find the artist with the most number of albums and output |
37 | | -the result of that. |
38 | | - |
39 | | -When done remove the database file and the downloaded content. |
40 | | -``` |
41 | | - |
42 | | -```shell |
43 | | -$ gptscript ./example.gpt |
44 | | -``` |
45 | | - |
46 | | -``` |
47 | | -OUTPUT: |
48 | | -
|
49 | | -The artist with the most number of albums in the database is Iron Maiden, with a total |
50 | | -of 21 albums. |
| 14 | +### Getting started |
| 15 | +MacOS and Linux: |
51 | 16 | ``` |
52 | | - |
53 | | -## Quick Start |
54 | | - |
55 | | -### 1. Install the latest release |
56 | | - |
57 | | -#### Homebrew (macOS and Linux) |
58 | | - |
59 | | -```shell |
60 | | -brew install gptscript-ai/tap/gptscript |
| 17 | +brew install gptscript-ai/tap/gptscript |
| 18 | +gptscript github.com/gptscript-ai/llm-basics-demo |
61 | 19 | ``` |
62 | 20 |
|
63 | | -#### Install Script (macOS and Linux): |
64 | | - |
65 | | -```shell |
66 | | -curl https://get.gptscript.ai/install.sh | sh |
| 21 | +Windows: |
67 | 22 | ``` |
68 | | - |
69 | | -### Scoop (Windows) |
70 | | - |
71 | | -```shell |
72 | | -scoop bucket add extras # If 'extras' is not already enabled |
73 | | -scoop install gptscript |
74 | | -``` |
75 | | - |
76 | | -#### WinGet (Windows) |
77 | | - |
78 | | -```shell |
79 | 23 | winget install gptscript-ai.gptscript |
| 24 | +gptscript github.com/gptscript-ai/llm-basics-demo |
80 | 25 | ``` |
81 | 26 |
|
82 | | -#### Manually |
83 | | - |
84 | | -Download and install the archive for your platform and architecture from the [releases page](https://github.com/gptscript-ai/gptscript/releases). |
85 | | - |
86 | | -### 2. Get an API key from [OpenAI](https://platform.openai.com/api-keys). |
87 | | - |
88 | | -#### macOS and Linux |
89 | | - |
90 | | -```shell |
91 | | -export OPENAI_API_KEY="your-api-key" |
92 | | -``` |
93 | | - |
94 | | -#### Windows |
95 | | - |
96 | | -```powershell |
97 | | -$env:OPENAI_API_KEY = 'your-api-key' |
98 | | -``` |
99 | | - |
100 | | -### 3. Run Hello World |
101 | | - |
102 | | -```shell |
103 | | -gptscript https://get.gptscript.ai/echo.gpt --input 'Hello, World!' |
104 | | -``` |
105 | | - |
106 | | -``` |
107 | | -OUTPUT: |
108 | | -
|
109 | | -Hello, World! |
110 | | -``` |
111 | | - |
112 | | -The model used by default is `gpt-4o` and you must have access to that model in your OpenAI account. |
113 | | - |
114 | | -### 4. Extra Credit: Examples and Run Debugging UI |
115 | | - |
116 | | -Clone examples and run debugging UI |
117 | | - |
118 | | -```shell |
119 | | -git clone https://github.com/gptscript-ai/gptscript |
120 | | -cd gptscript/examples |
121 | | - |
122 | | -# Run the debugging UI |
123 | | -gptscript --server |
124 | | -``` |
125 | | - |
126 | | -## How it works |
127 | | - |
128 | | -**_GPTScript is composed of tools._** Each tool performs a series of actions similar to a function. Tools have available |
129 | | -to them other tools that can be invoked similar to a function call. While similar to a function, the tools are |
130 | | -primarily implemented with a natural language prompt. **_The interaction of the tools is determined by the AI model_**, |
131 | | -the model determines if the tool needs to be invoked and what arguments to pass. Tools are intended to be implemented |
132 | | -with a natural language prompt but can also be implemented with a command or HTTP call. |
133 | | - |
134 | | -### Example |
135 | | - |
136 | | -Below are two tool definitions, separated by `---`. The first tool does not require a name or description, but |
137 | | -every tool after name and description are required. The first tool, has the parameter `tools: bob` meaning that the tool named `bob` is available to be called if needed. |
138 | | - |
139 | | -```yaml |
140 | | -tools: bob |
141 | | - |
142 | | -Ask Bob how he is doing and let me know exactly what he said. |
143 | | - |
144 | | ---- |
145 | | -name: bob |
146 | | -description: I'm Bob, a friendly guy. |
147 | | -args: question: The question to ask Bob. |
148 | | - |
149 | | -When asked how I am doing, respond with "Thanks for asking "${question}", I'm doing great fellow friendly AI tool!" |
150 | | -``` |
151 | | - |
152 | | -Put the above content in a file named `bob.gpt` and run the following command: |
153 | | - |
154 | | -```shell |
155 | | -$ gptscript bob.gpt |
156 | | -``` |
157 | | - |
158 | | -``` |
159 | | -OUTPUT: |
160 | | -
|
161 | | -Bob said, "Thanks for asking 'How are you doing?', I'm doing great fellow friendly AI tool!" |
162 | | -``` |
163 | | - |
164 | | -Tools can be implemented by invoking a program instead of a natural language prompt. The below |
165 | | -example is the same as the previous example but implements Bob using python. |
166 | | - |
167 | | -```yaml |
168 | | -Tools: bob |
169 | | - |
170 | | -Ask Bob how he is doing and let me know exactly what he said. |
171 | | - |
172 | | ---- |
173 | | -Name: bob |
174 | | -Description: I'm Bob, a friendly guy. |
175 | | -Args: question: The question to ask Bob. |
176 | | - |
177 | | -#!python3 |
178 | | - |
179 | | -import os |
180 | | - |
181 | | -print(f"Thanks for asking {os.environ['question']}, I'm doing great fellow friendly AI tool!") |
182 | | -``` |
183 | | - |
184 | | -With these basic building blocks you can create complex scripts with AI interacting with AI, your local system, data, |
185 | | -or external services. |
186 | | - |
187 | | -## GPT File Reference |
188 | | - |
189 | | -### Extension |
190 | | - |
191 | | -GPTScript files use the `.gpt` extension by convention. |
192 | | - |
193 | | -### File Structure |
194 | | - |
195 | | -A GPTScript file has one or more tools in the file. Each tool is separated by three dashes `---` alone on a line. |
196 | | - |
197 | | -```yaml |
198 | | -Name: tool1 |
199 | | -Description: This is tool1 |
200 | | - |
201 | | -Do sample tool stuff. |
202 | | - |
203 | | ---- |
204 | | -Name: tool2 |
205 | | -Description: This is tool2 |
206 | | - |
207 | | -Do more sample tool stuff. |
208 | | -``` |
209 | | - |
210 | | -### Tool Definition |
211 | | - |
212 | | -A tool starts with a preamble that defines the tool's name, description, args, available tools and additional parameters. |
213 | | -The preamble is followed by the tool's body, which contains the instructions for the tool. Comments in |
214 | | -the preamble are lines starting with `#` and are ignored by the parser. Comments are not really encouraged |
215 | | -as the text is typically more useful in the description, argument descriptions or instructions. |
216 | | - |
217 | | -```yaml |
218 | | -Name: tool-name |
219 | | -# This is a comment in the preamble. |
220 | | -Description: Tool description |
221 | | -# This tool can invoke tool1 or tool2 if needed |
222 | | -Tools: tool1, tool2 |
223 | | -Args: arg1: The description of arg1 |
224 | | - |
225 | | -Tool instructions go here. |
226 | | -``` |
227 | | - |
228 | | -#### Tool Parameters |
229 | | - |
230 | | -Tool parameters are key-value pairs defined at the beginning of a tool block, before any instructional text. They are specified in the format `key: value`. The parser recognizes the following keys (case-insensitive and spaces are ignored): |
231 | | - |
232 | | - |
233 | | -| Key | Description | |
234 | | -|------------------|-----------------------------------------------------------------------------------------------------------------------------------------| |
235 | | -| `Name` | The name of the tool. | |
236 | | -| `Model Name` | The OpenAI model to use, by default it uses "gpt-4-turbo" | |
237 | | -| `Description` | The description of the tool. It is important that this properly describes the tool's purpose as the description is used by the LLM. | |
238 | | -| `Internal Prompt`| Setting this to `false` will disable the built-in system prompt for this tool. | |
239 | | -| `Tools` | A comma-separated list of tools that are available to be called by this tool. | |
240 | | -| `Args` | Arguments for the tool. Each argument is defined in the format `arg-name: description`. | |
241 | | -| `Max Tokens` | Set to a number if you wish to limit the maximum number of tokens that can be generated by the LLM. | |
242 | | -| `JSON Response` | Setting to `true` will cause the LLM to respond in a JSON format. If you set true you must also include instructions in the tool. | |
243 | | -| `Temperature` | A floating-point number representing the temperature parameter. By default, the temperature is 0. Set to a higher number for more creativity. | |
244 | | - |
245 | | - |
246 | | -#### Tool Body |
247 | | - |
248 | | -The tool body contains the instructions for the tool which can be a natural language prompt or |
249 | | -a command to execute. Commands must start with `#!` followed by the interpreter (e.g. `#!/bin/bash`, `#!python3`) |
250 | | -a text that will be placed in a file and passed to the interpreter. Arguments can be references in the instructions |
251 | | -using the format `${arg1}`. |
252 | | - |
253 | | -```yaml |
254 | | -name: echo-ai |
255 | | -description: A tool that echos the input |
256 | | -args: input: The input |
257 | | - |
258 | | -Just return only "${input}" |
259 | | - |
260 | | ---- |
261 | | -name: echo-command |
262 | | -description: A tool that echos the input |
263 | | -args: input: The input |
264 | | - |
265 | | -#!/bin/bash |
266 | | - |
267 | | -echo "${input}" |
268 | | -``` |
269 | | - |
270 | | -## Built in Tools |
271 | | - |
272 | | -There are several built in tools to do basic things like read/write files, download http content and execute commands. |
273 | | -Run `gptscript --list-tools` to list all the built-in tools. |
274 | | - |
275 | | -## Examples |
276 | | - |
277 | | -For more examples check out the [examples](examples) directory. |
| 27 | +A few notes: |
| 28 | +- You'll need an [OpenAI API key](https://help.openai.com/en/articles/4936850-where-do-i-find-my-openai-api-key) |
| 29 | +- On Windows, after installing gptscript you may need to restart your terminal for the changes to take effect |
| 30 | +- The above script is a simple chat-based assistant. You can ask it questions and it will answer to the best of its ability. |
278 | 31 |
|
279 | 32 | ## Community |
280 | 33 |
|
|
0 commit comments