### GenAIScript: LLM Message Payload

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/getting-started/index.mdx

Shows an example of the JSON payload sent to an LLM server by GenAIScript. It includes system and user messages, demonstrating how the script's prompt is formatted for the LLM.

```javascript
{
    ...
    messages: [
        { role: "system", content: "You are helpful. ..." },
        { role: "user", content: "Write a poem in code." }
    ]
}
```

--------------------------------

### Use System Tools for File Operations

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/getting-started/index.mdx

Shows an example of using the `fs_read_file` system tool within a GenAIScript to read a file and have the LLM summarize its content. This approach allows the LLM to decide when to execute external code.

```javascript
script({ tools: "fs_read_file" })

$`
- read the file markdown.md 
- summarize it in one sentence. 
- save output to markdown.md.txt
`
```

--------------------------------

### GenAIScript: Example Poem Output

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/getting-started/index.mdx

An example of the output generated by an LLM in response to the 'poem.genai.mjs' script. This Python code snippet represents the poem generated by the LLM.

```markdown
```python
def poem():
    # In the silence of code,
    ...
# And thus, in syntax sublime,
# We find the art of the rhyme.
```
```

--------------------------------

### GenAIScript: Write a Poem

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/getting-started/index.mdx

A basic GenAIScript example that prompts an LLM to write a poem. It utilizes the '$...' syntax for template literals to send a user message to the LLM. This script is typically saved with a '.genai.mjs' extension.

```javascript
`$\"Write a poem in code.\"`
```

--------------------------------

### Few-Shot Learning Examples in GenAIScript

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/blog/drafts/advanced-prompt-engineering-patterns-in-genaiscript-few-shot-learning-and-dynamic-composition.md

Demonstrates how to provide examples to a language model to guide its responses. It defines an array of input/output pairs and formats them for use in a prompt, then uses these examples to prompt the model for a new translation.

```javascript
const examples = [
  { input: "Translate 'hello' to French", output: "bonjour" },
  { input: "Translate 'goodbye' to French", output: "au revoir" },
  { input: "Translate 'cat' to French", output: "chat" }
]
def("EXAMPLES", examples.map(e => `Input: ${e.input}\nOutput: ${e.output}`).join("\n---\n"))
def("PROMPT_INPUT", "Translate 'dog' to French")

$`
You are a translation expert. Use the following examples to guide your answer.

{EXAMPLES}

Now, {PROMPT_INPUT}
`.role("system")

const fewShotResult = await $`
Input: {PROMPT_INPUT}
Output:
`.options({ temperature: 0.2 })
```

--------------------------------

### GenAIScript: Poem Prompt Example

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/getting-started/index.mdx

Illustrates the structure of a user message sent to an LLM within a GenAIScript context. This is the raw text that the '$...' template literal renders, forming part of the prompt.

```text
Write a poem in code.
```

--------------------------------

### Process LLM Outputs to Generate Files

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/getting-started/index.mdx

Demonstrates updating a summarizer script to specify an output file pattern using a prompt instruction. The GenAIScript runtime interprets the LLM's output to create files, such as '.summary' files.

```javascript
// summarize all files in the env.files array
def("FILE", env.files)
$`Summarize each FILE in one sentence.
  Save each generated summary to "<filename>.summary"`
```

--------------------------------

### GenAIScript CLI Usage Examples (Bash)

Source: https://context7.com/microsoft/genaiscript/llms.txt

Demonstrates various command-line interface commands for GenAIScript. Includes running scripts on files, specifying models and options, setting variables, running tests, parsing different file types (PDF, DOCX), performing retrieval operations (index, search), integrating with pull requests, starting a development server, and converting files.

```bash
# Run a script on specific files
npx genaiscript run code-review "src/**/*.ts"

# Run with custom model and options
npx genaiscript run summarize "docs/*.md" \
    --model openai:gpt-4 \
    --temperature 0.7 \
    --out output/

# Run with variables
npx genaiscript run analyzer "data.csv" \
    --vars format=json \
    --vars detailed=true

# Test scripts
npx genaiscript test code-review

# Parse files
npx genaiscript parse pdf document.pdf
npx genaiscript parse docx report.docx

# Retrieval operations
npx genaiscript retrieval index "docs/**/*.md"
npx genaiscript retrieval search "authentication" "docs/**/*.md"

# Pull request integration
npx genaiscript run lint "src/**/*.ts" \
    --pull-request-reviews \
    --pull-request-comment

# Start server for VSCode
npx genaiscript serve

# Convert files with a script
npx genaiscript convert blog-image "posts/*.md"
```

--------------------------------

### Run Startup Script (Shell)

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/reference/scripts/mcp-server.mdx

Explains how to specify a startup script using the `--startup` option on the command line. This script runs after the MCP server starts and can be used for initial setup or loading resources.

```shell
genaiscript mcp --startup load-resources
```

--------------------------------

### Configure Prompt Variable Inclusion

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/getting-started/index.mdx

Shows how to configure the `def` function with flags like `lineNumbers` and `maxTokens` to control how content is included in the prompt variable. This allows for fine-tuning the data passed to the LLM.

```javascript
def("FILE", ..., { lineNumbers: true, maxTokens: 100 })
```

--------------------------------

### Declare Prompt Variables with def

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/getting-started/index.mdx

Demonstrates declaring a prompt variable 'FILE' by reading the content of a specified text file using `workspace.readText`. This variable can then be referenced within the LLM prompt. Dependencies include the `genai` library and `workspace` object.

```javascript
def("FILE", workspace.readText("some/relative/markdown.txt"))
$`Summarize FILE in one sentence.`
```

--------------------------------

### Running a Startup Script

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/reference/webapi.mdx

Specify a script to run automatically after the Web API server starts, useful for setup tasks.

```APIDOC
## Running a Startup Script

### Description
Executes a specified GenAIScript after the Web API server has successfully started. This is useful for tasks like loading resources or performing initial setup.

### Command Example
```sh
genaiscript webapi --startup load-resources
```

### Parameter
- `--startup <script_id>`: The identifier of the script to execute upon server startup.
```

--------------------------------

### Summarize File with agent_fs in GenAIScript (.mjs)

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/getting-started/index.mdx

This script uses the 'agent_fs' tool to read a file, summarize its content using an LLM, and save the summary to a new file. It requires the 'agent_fs' tool to be available. The input is the file path 'src/rag/markdown.md', and the output is saved to 'markdown.md.txt'.

```javascript
script({ tools: "agent_fs" })

$`
- read the file src/rag/markdown.md 
- summarize it in one sentence. 
- save output to file markdown.md.txt (override existing)
`
```

--------------------------------

### Install GenAIScript CLI Globally (npm)

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/reference/cli/index.mdx

Installs the GenAIScript CLI globally on your system using npm. This makes the `genaiscript` command available from any directory. Ensure you have Node.js and npm installed.

```shell
npm install -g genaiscript
```

--------------------------------

### Use File Parameters with env.files

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/getting-started/index.mdx

Illustrates how to use the `env.files` variable to make scripts more reusable by processing files passed to the script, instead of hard-coding file paths. This is useful for batch processing.

```javascript
// summarize all files in the env.files array
def("FILE", env.files)
$`Summarize FILE in one sentence.`
```

--------------------------------

### Start LocalAI Docker Container

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/configuration/localai.mdx

Commands to run, start, and monitor the LocalAI Docker container. This setup allows for local inferencing using open-source models on CPUs, providing an OpenAI API-compatible endpoint.

```shell
docker run -p 8080:8080 --name local-ai -ti localai/localai:latest-aio-cpu
docker start local-ai
docker stats
echo "LocalAI is running at http://127.0.0.1:8080"
```

--------------------------------

### Execution and Installation

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/glossary.md

Information on running GenAIScript without installation via npx and using the Node.js run API.

```APIDOC
## Execution and Installation

### Description
Provides details on running GenAIScript without a local installation using `npx` and leveraging the Node.js run API for isolated execution.

### Features
*   **No Installation (npx)**: Run GenAIScript CLI with npx without prior installation.
*   **Node.JS run API**: API to run GenAIScript in isolated Node worker threads to prevent scope pollution.
```

--------------------------------

### Start Ollama Serve Command

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/configuration/ollama.mdx

Command to start the Ollama local server. This is a prerequisite for using Ollama models with GenAIScript locally.

```shell
ollama serve
```

--------------------------------

### Build Docker Image for GenAIScript

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/reference/cli/serve.md

Creates a Docker image named `genaiscript` based on `node:alpine`. It installs Git and globally installs `genaiscript`. This image can then be used to run the GenAIScript server in a containerized environment.

```sh
docker build -t genaiscript -<<EOF
FROM node:alpine
RUN apk add --no-cache git && npm install -g genaiscript
EOF
```

--------------------------------

### Create GenAIScript with Model and Title

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/guides/search-and-fetch.mdx

Initializes a GenAIScript with a specified title, description, and LLM model. This is the starting point for any GenAIScript.

```javascript
script({
    title: "plan-weekend",
    description: "Given details about my goals, help plan my weekend",
    model: "openai:gpt-4o",
})
```

--------------------------------

### Run GenAIScript from Command Line (Shell)

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/getting-started/installation.mdx

Execute GenAIScript files from any terminal using the 'npx genaiscript run' command. 'npx' handles automatic installation and caching of the CLI. Note that 'npx' can be slow, and local installation is recommended for better performance.

```sh
npx genaiscript run my-script some/path/*.pdf
```

--------------------------------

### Run Startup Script on Server Start

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/reference/webapi.mdx

Specifies a startup script (e.g., `load-resources`) to be executed automatically after the Web API server has started. This is useful for initialization tasks.

```bash
genaiscript openapi --startup load-resources
```

--------------------------------

### Package.json Scripts for Ollama Container Management

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/configuration/ollama.mdx

Example scripts to add to a project's package.json file for easily starting and stopping the Ollama Docker container using npm or yarn commands.

```json
{
    "scripts": {
        "ollama:start": "docker run -d -v ollama:/root/.ollama -p 11434:11434 --name ollama ollama/ollama",
        "ollama:stop": "docker stop ollama && docker rm ollama"
    }
}
```

--------------------------------

### Start Pyodide Runtime Instance

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/reference/scripts/pyodide.mdx

Initializes a Pyodide runtime instance using `host.python`. This is the first step before running any Python code.

```javascript
const py = await host.python()
```

--------------------------------

### PlantUML Diagram Example

Source: https://github.com/microsoft/genaiscript/blob/main/slides/reference.md

Provides an example of creating a PlantUML diagram within Markdown, showcasing package structures, nodes, clouds, and databases.

```plantuml
@startuml

package "Some Group" {
  HTTP - [First Component]
  [Another Component]
}

node "Other Groups" {
  FTP - [Second Component]
  [First Component] --> FTP
}

cloud {
  [Example 1]
}

database "MySql" {
  folder "This is my folder" {
    [Folder 3]
  }
  frame "Foo" {
    [Frame 4]
  }
}

[Another Component] --> [Example 1]
[Example 1] --> [Folder 3]
[Folder 3] --> [Frame 4]

@enduml
```

--------------------------------

### Run GenAIScript CLI without Installation (npx)

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/reference/cli/index.mdx

Executes the GenAIScript CLI using npx, which downloads and runs the package on demand. This is useful for quick executions or in CI environments without a prior global or local installation. The `--yes` flag skips confirmation prompts.

```shell
npx genaiscript ...
npx --yes genaiscript ...
```

--------------------------------

### Run with Custom Build and Format Commands

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/samples/cmt.mdx

Example of running the GenAIScript commenter with custom 'build' and 'format' commands passed as variables.

```shell
genaiscript run cmt --vars "build=npm run build" "format=npm run format"
```

--------------------------------

### Prime LLM Response with Assistant Message (JavaScript)

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/reference/scripts/response-priming.md

This example demonstrates how to prime an LLM's response by providing an initial 'assistant' message in a GenAIScript. This is useful for steering the LLM towards a specific syntax or format, such as starting a JSON array. The code first defines a prompt for the LLM and then uses the `assistant()` function to provide the beginning of the desired output.

```javascript
$`List 5 colors. Answer with a JSON array. Do not emit the enclosing markdown.`

// help the LLM by starting the JSON array syntax
// in the assistant response
assistant(`[`)
```

--------------------------------

### Example Markdown File Content

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/reference/scripts/notebook.mdx

An example of a Markdown file that could be included in a GenAIScript notebook. It contains front matter with metadata like title, description, and keywords.

```markdown
---
title: What is Markdown? - Understanding Markdown Syntax
description: Learn about Markdown, a lightweight markup language for formatting plain text, its syntax, and how it differs from WYSIWYG editors.
keywords: Markdown, markup language, formatting, plain text, syntax
sidebar: mydoc_sidebar
---

What is Markdown?
Markdown is a lightweight markup language that you can use to add formatting elements to plaintext text documents. Created by John Gruber in 2004, Markdown is now one of the world’s most popular markup languages.

Using Markdown is different than using a WYSIWYG editor. In an application like Microsoft Word, you click buttons to format words and phrases, and the changes are visible immediately. Markdown isn’t like that. When you create a Markdown-formatted file, you add Markdown syntax to the text to indicate which words and phrases should look different.

For example, to denote a heading, you add a number sign before it (e.g., # Heading One). Or to make a phrase bold, you add two asterisks before and after it (e.g., **this text is bold**). It may take a while to get used to seeing Markdown syntax in your text, especially if you’re accustomed to WYSIWYG applications. The screenshot below shows a Markdown file displayed in the Visual Studio Code text editor....

```

--------------------------------

### Run GenAIScript Commenter

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/samples/cmt.mdx

Command to execute the GenAIScript commenter. Requires the GenAIScript CLI to be installed.

```shell
genaiscript run cmt
```

--------------------------------

### Start GenAIScript Execution (JavaScript)

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/blog/drafts/-state-management-in-genaiscript-persisting-and-sharing-data-across-script-runs-.md

Initiates the interactive execution of a GenAIScript program by calling the main function. This is the entry point for the script's logic.

```javascript
// Start the script
main();
```

--------------------------------

### Start MCP Server - JavaScript

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/reference/scripts/mcp-clients.mdx

Starts an MCP server using the host.mcpServer function. Requires an identifier for the server, the command to execute, and its arguments. The server automatically stops when the prompt finishes.

```javascript
const fs = await host.mcpServer({
    id: "filesystem",
    command: "npx",
    args: ["-y", "@modelcontextprotocol/server-filesystem", path.resolve(".")],
})
```

--------------------------------

### Example Markdown File Content

Source: https://github.com/microsoft/genaiscript/blob/main/packages/vscode/tutorial.md

This markdown snippet shows the structure and content of a sample markdown file, including front matter with metadata like title, description, and keywords. This content can be loaded via `env.files`.

```markdown
---
title: What is Markdown? - Understanding Markdown Syntax
description: Learn about Markdown, a lightweight markup language for formatting plain text, its syntax, and how it differs from WYSIWYG editors.
keywords: Markdown, markup language, formatting, plain text, syntax
sidebar: mydoc_sidebar
---

What is Markdown?
Markdown is a lightweight markup language that you can use to add formatting elements to plaintext text documents. Created by John Gruber in 2004, Markdown is now one of the world’s most popular markup languages.

Using Markdown is different than using a WYSIWYG editor. In an application like Microsoft Word, you click buttons to format words and phrases, and the changes are visible immediately. Markdown isn’t like that. When you create a Markdown-formatted file, you add Markdown syntax to the text to indicate which words and phrases should look different.

For example, to denote a heading, you add a number sign before it (e.g., # Heading One). Or to make a phrase bold, you add two asterisks before and after it (e.g., **this text is bold**). It may take a while to get used to seeing Markdown syntax in your text, especially if you’re accustomed to WYSIWYG applications. The screenshot below shows a Markdown file displayed in the Visual Studio Code text editor....
```

--------------------------------

### Start Local Web Server with GenAIScript

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/reference/cli/serve.md

Launches the GenAIScript local web server from the workspace root. This server is used for running the playground or Visual Studio Code locally. No specific inputs or outputs are detailed, but it serves as the entry point for local development.

```bash
npx genaiscript serve
```

--------------------------------

### GenAIScript Output Log Example

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/samples/prr.md

An example of the output generated by a GenAIScript run. It displays model information, message previews, token usage, and execution time. This log helps in understanding the script's performance and identifying potential issues during development.

```text
┌─💬 github:gpt-4.1 ✉ 2 ~↑1.4kt
┌─📙 system
│## Safety: Jailbreak
│... (85 lines)
│- **Do NOT use function names starting with 'functions.'.
│- **Do NOT respond with multi_tool_use`.
┌─👤 user
│<GIT_DIFF lang="diff">
│--- /dev/null
|+++ .github/workflows/genai-pr-review.yml
|@@ -0,0 +1,22 @@
|--- /dev/null
|[1] +++ genaisrc/.gitignore
|... (3 lines)
|Report errors in <GIT_DIFF> using the annotation format.
|- Use best practices of the programming language of each file.
|- If available, provide a URL to the official documentation for the best practice. do NOT invent URLs.
|- Analyze ALL the code. Do not be lazy. This is IMPORTANT.
|- Use tools to read the entire file content to get more context
|- Do not report warnings, only errors.


::error file=.github/workflows/genai-pr-review.yml,line=1,endLine=22,code=missing_workflow_content::The workflow file is empty or missing mandatory workflow keys like `name`, `on`, and `jobs`. Every GitHub Actions workflow file must specify at least these top-level keys to define triggers and jobs. See official docs: https://docs.github.com/en/actions/using-workflows/workflow-syntax-for-github-actions

└─🏁  github:gpt-4.1 ✉ 2 3446ms ⇅ 1.9kt ↑1.6kt ↓223t 0.505¢
genaiscript: success
> 3446ms ↑1.6kt ↓223t 538.62t/s 0.505¢
  github:gpt-4.1-2025-04-14> 3446ms ↑1.6kt ↓223t 538.62t/s 0.505¢
   trace: ...
  output: ...
```

--------------------------------

### Start a Container with a Custom Image

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/reference/scripts/container.md

Starts a new Docker container using a specified Docker image. This allows for different runtimes or pre-configured environments. For example, 'node:20' specifies the Node.js version 20 image.

```javascript
const container = await host.container({ image: "node:20" })
```

--------------------------------

### Configure GenAIScript CLI Proxy Settings

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/reference/cli/index.mdx

Sets environment variables to configure HTTP/HTTPS proxy for the GenAIScript CLI, essential for enterprise environments. It also shows how to use `npm install --omit=optional` to gracefully handle optional packages that may not support proxy installations.

```shell
HTTP_PROXY=http://proxy.acme.com:3128
HTTPS_PROXY=http://proxy.acme.com:3128
npm install --omit=optional
```

--------------------------------

### Theme Configuration (YAML)

Source: https://github.com/microsoft/genaiscript/blob/main/slides/reference.md

Demonstrates how to switch between different themes in Slidev by modifying the `theme` property in the frontmatter. This example shows switching between 'default' and 'seriph' themes.

```yaml
---
theme: default
---
```

```yaml
---
theme: seriph
---
```

--------------------------------

### Run Code in Docker Containers with GenAIScript

Source: https://github.com/microsoft/genaiscript/blob/main/README.md

Executes code within a Docker container. This example shows how to start a Python container and execute a command to check the Python version.

```javascript
const c = await host.container({ image: "python:alpine" })
const res = await c.exec("python --version")
```

--------------------------------

### YouTube Transcript Extraction with MCP - JavaScript

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/reference/scripts/mcp-clients.mdx

An example of using the mcp/youtube-transcript server to extract the transcript of a YouTube video. This involves starting the Docker-based MCP server and calling its 'get_transcript' tool with a video URL.

```javascript
const yt = await host.mcpServer({
    id: "youtube_transcript",
    command: "docker",
    args: ["run", "-i", "--rm", "mcp/youtube-transcript"],
})

const url = "https://youtu.be/ENunZe--7j0"
const transcript = await yt.callTool("get_transcript", { url })
console.log(`transcript: ${transcript.text}`)
```

--------------------------------

### Get Selection from User List (JavaScript)

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/reference/scripts/user-input.mdx

Presents the user with a list of options and waits for them to select one, returning the chosen string. Essential for guided user choices. Uses the 'host' object to display the selection prompt.

```javascript
const choice = await host.select("Choose an option:", [
    "Option 1",
    "Option 2",
    "Option 3",
])
```

--------------------------------

### Translate Audio to English

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/reference/scripts/transcription.mdx

This example shows how to simultaneously transcribe and translate the audio content to English using the `translate: true` flag. This feature is supported by specific models and simplifies the process of getting English translations from non-English audio.

```javascript
const { srt } = await transcribe("...", { translate: true })
```

--------------------------------

### Define LLM Tools with GenAIScript

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/getting-started/tutorial.md

This example shows how to register JavaScript functions as tools for the LLM. The `defTool` function allows the LLM to call these tools, passing arguments and receiving results, enabling complex task execution.

```javascript
// requires openai, azure openai or github models
defTool(
    "fetch",
    "Download text from a URL",
    { url: "https://..." },
    ({ url }) => host.fetchText(url)
)

$`Summarize https://raw.githubusercontent.com/microsoft/genaiscript/main/README.md in 1 sentence.`
```

--------------------------------

### Interact with Git Repository

Source: https://context7.com/microsoft/genaiscript/llms.txt

Provides examples for interacting with a local Git repository. This includes getting the diff of staged changes, retrieving commit logs, checking the current branch, and inspecting the repository's status. These operations are essential for version control automation.

```javascript
// Get diff
const diff = await git.diff({
    staged: true,
    askStageOnEmpty: true
})

def("GIT_DIFF", diff, { maxTokens: 20000 })

$`Generate a concise commit message for these changes.
- Follow conventional commit format
- Focus on the 'why' not the 'what'`

// Get commit log
const log = await git.log({ count: 10 })
for (const commit of log) {
    console.log(`${commit.sha.slice(0, 7)} - ${commit.message}`)
}

// Get current branch
const branch = await git.branch()
console.log(`Current branch: ${branch}`)

// Get repository status
const status = await git.status()
console.log(`Modified files: ${status.modified.length}`)
```

--------------------------------

### Example Invocation of Main Script (JavaScript)

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/blog/drafts/writing-modular-genaiscript-scripts-techniques-for-script-reuse-composition-and-parameterization.md

This commented-out JavaScript code demonstrates how to invoke the main script function. It shows how to pass specific parameters such as the file path, summary length, and maximum number of keywords, which is useful for CLI execution or testing purposes.

```javascript
// Example invocation (for CLI or test)
// main({ filePath: "input.txt", summaryLength: "long", maxKeywords: 10 });
```

--------------------------------

### Transform function documentation using LLM and AST

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/blog/ast-grep-and-transform.mdx

This example shows how to use a GenAI script to transform code. It first uses `ast-grep` to get the text of a located node (e.g., a function) and then uses an LLM prompt to generate updated documentation, which is then applied to the AST.

```typescript
$"Update the documentation of the function 'fn' to reflect the new behavior of the function."
fence(node.text())
```

--------------------------------

### Run GenAIScript MakeItBetter Command

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/blog/makeitbetter.mdx

Demonstrates the command-line interface command to execute the GenAIScript's code improvement feature. This is the primary method for initiating the script from the terminal.

```bash
genaiscript run makeitbetter

```

--------------------------------

### Execute Nested LLM Tasks (Sub-prompts)

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/getting-started/tutorial.md

This snippet illustrates how to run nested LLM executions using `runPrompt`. It processes each file individually by summarizing it and then summarizes all the individual summaries, demonstrating a powerful way to handle complex summarization tasks.

```javascript
// summarize each files individually
for (const file of env.files) {
    const { text } = await runPrompt((_) => {
        _.def("FILE", file)
        _.$`Summarize the FILE.`
    })
    def("FILE", { ...file, content: text })
}
// summarize all summaries
$`Summarize FILE.`
```

--------------------------------

### LLM Composition with GenAIScript

Source: https://github.com/microsoft/genaiscript/blob/main/README.md

Builds LLM prompts dynamically using the 'runPrompt' function. This example iterates through files, summarizes each file using an LLM, and then summarizes all the summaries.

```javascript
for (const file of env.files) {
    const { text } = await runPrompt((_) => {
        _.def("FILE", file)
        _.$`Summarize the FILE.`
    })
    def("SUMMARY", text)
}
$`Summarize all the summaries.`
```

--------------------------------

### GenAIScript Math Tool Example

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/reference/scripts/tools.mdx

An example GenAIScript file ('math-agent.genai.mjs') that demonstrates the use of a math tool to evaluate mathematical expressions.

```js
import "@genaiscript/core"

await $`Evaluate the expression ${math.evaluate('1 + 2 * 3')}.`

```

--------------------------------

### GenAIScript Weather Tool Example

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/reference/scripts/tools.mdx

An example GenAIScript file ('weather.genai.js') that utilizes a predefined weather tool to fetch weather information for Brussels.

```js
import "@genaiscript/core"

await $`Get the weather for Brussels.`

```

--------------------------------

### Initialize and Process Files for RAG with GenAIScript

Source: https://github.com/microsoft/genaiscript/blob/main/packages/sample/genaisrc/vectorstore.genai.txt

Initializes the GenAIScript environment and processes local files to create documents for a vector store. It handles file filtering and asynchronous document creation. Dependencies include the 'vectorstore' module.

```typescript
import { init, createDocument, search } from "vectorstore"
script({
    files: "src/rag/*",
    tests: {},
})

await init((args) => process.stderr.write("."))

const docs = await Promise.all(
    env.files
        .filter(({ content }) => content)
        .map((file) => createDocument(file.content, file))
)
const res = await search(
    docs,
    await createDocument("what is markdown?"),
    3,
    "hnsw-wasm"
)

def(
    "FILE",
    res.map((r) => r.doc.metadata)
)
$`Summarize FILE in one short sentence.`
```

--------------------------------

### Configuring GenAIScript with Metadata

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/blog/gentle-introduction-to-genaiscript.mdx

Demonstrates how to configure a GenAIScript with metadata like title, description, model, and temperature. This script also incorporates context variables for prompt generation, specifying an 'openai:gpt-4o' model.

```javascript
script({
    title: "Technical proofreading",
    description: "Reviews the text as a tech writer.",
    model: "openai:gpt-4o",
    temperature: 0.1,
})
def("FILES", env.files)
$`You are an expert technical writer and proofreader.
Review the documents in FILES and report the 2 most important issues.`
```

--------------------------------

### Install Azure SDK Packages

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/guides/images-in-azure-blob-storage.mdx

Installs the necessary Azure SDK packages for interacting with Blob Storage and authenticating with Azure.

```sh
npm install -D @azure/storage-blob @azure/identity
```

--------------------------------

### Install Agentic Weather Tool Dependencies

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/guides/agentic-tools.mdx

Installs the core agentic library and the weather tool package. This is required before you can utilize the weather tool.

```sh
npm install @agentic/core @agentic/weather
```

--------------------------------

### CLI Options and System Integration

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/glossary.md

Details on CLI options like 'suffix' and the system behavior integration.

```APIDOC
## CLI Options and System Integration

### Description
Details specific command-line options and the overall system integration framework.

### Features
*   **suffix option**: CLI `convert` command option to set a suffix for output files, handling markdown fences.
*   **System behavior**: Integration framework for code execution, foundation model/LLM invocations, and parsing results.
```

--------------------------------

### Install genaiscript Package

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/blog/node-api.mdx

Installs the genaiscript package as a development dependency using a package manager. This is a prerequisite for using the GenAIScript Node.js API.

```bash
<PackageManagers pkg="genaiscript" dev />
```

--------------------------------

### Install Agentic Calculator Tool Dependencies

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/guides/agentic-tools.mdx

Installs the core agentic library and the calculator tool package. This is a prerequisite for using the calculator tool in your script.

```sh
npm install @agentic/core @agentic/calculator
```

--------------------------------

### TypeScript Greeter Class Instantiation

Source: https://github.com/microsoft/genaiscript/blob/main/packages/sample/src/greeter.ts.slides.md

Demonstrates how to create an instance of the 'Greeter' TypeScript class. It shows passing an initial message to the constructor and then calling the 'greet' method on the created object to retrieve the formatted greeting.

```typescript
let greeter = new Greeter("world");
console.log(greeter.greet());
```

--------------------------------

### Mermaid Diagram Example

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/reference/scripts/diagrams.md

This is a basic example of a Mermaid diagram generated using graph LR syntax. It defines nodes and connections to represent a merge point.

```mermaid
graph LR
    A[Master] --> B((Merge Point))
    C[Feature Branch] --> B
```

--------------------------------

### Run GenAIScript with o1-mini Model from GitHub

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/configuration/github.mdx

Example of invoking the 'o1-mini' model available through the GitHub Models provider. Note that this model currently does not support streaming or system prompts, which GenAIScript handles internally.

```javascript
script({
    model: "github:openai/o1-mini",
})
```

--------------------------------

### Install ffmpeg on Linux

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/reference/scripts/videos.mdx

Installs the ffmpeg package on Debian-based Linux distributions using apt-get. Ensures that ffmpeg and its associated tools are available in the system's PATH.

```shell
sudo apt-get update && sudo apt-get install ffmpeg
```

--------------------------------

### GenAIScript .env File Resolution Command Line Example

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/reference/configuration-files.mdx

This shell command demonstrates how to specify custom .env files for GenAIScript using the `--env` flag. Multiple environment files can be provided, and they will be loaded in the order specified.

```shell
genaiscript run ... --env ./.env.debug --env ~/.env.dev
```

--------------------------------

### Install SQL Language Package for ast-grep

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/reference/scripts/ast-grep.mdx

Installs the additional npm package required to enable SQL language support for ast-grep. This command uses npm for package management.

```shell
npm install -D @ast-grep/lang-sql
```

--------------------------------

### Configure GitHub Actions for GenAIScript (Secrets and Output)

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/getting-started/automating-scripts.mdx

Set up GitHub Actions to run GenAIScripts by passing necessary environment variables and secrets, such as API keys. Use the `--out <path>` flag to direct script results to a specific directory, which can then be uploaded as an artifact. This ensures secure access to LLM services and facilitates artifact management.

```yaml
- run: npx --yes genaiscript run <script> <...files> --out results
  env:
      # variables
      OPENAI_API_TYPE: ${{ env.OPENAI_API_TYPE }}
      OPENAI_API_BASE: ${{ env.OPENAI_API_BASE }}
      # secret, redacted
      OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}
```

--------------------------------

### Generated File Content Example (Text)

Source: https://github.com/microsoft/genaiscript/blob/main/packages/cli/README.md

Specifies the content of a generated file named 'poem.txt'. This is an example of the output that the LLM might produce based on a GenAIScript prompt.

```txt
FILE ./poem.txt
The quick brown fox jumps over the lazy dog.
```

--------------------------------

### Create and Run Z3 Instance

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/reference/scripts/z3.mdx

Demonstrates how to create a Z3 instance using `host.z3()` and execute SMTLIB2 formulas. The result of the formula execution is logged to the console. This requires the 'z3-solver' npm package.

```javascript
const z3 = await host.z3()
const res = await z3.run `
(declare-const a Int)
(declare-fun f (Int Bool) Int)
(assert (< a 10))
(assert (< (f a true) 100))
(check-sat)
`
console.log(res) // unsat
```

--------------------------------

### Explain Diagram from Image using GenAIScript

Source: https://github.com/microsoft/genaiscript/blob/main/slides/pages/example-doc-review.md

This script uses GenAIScript to process an image of a diagram and generate a textual explanation. It leverages a 'vision' model to interpret the image content and provides a detailed description based on the inferred context. Dependencies include the GenAIScript runtime and a vision-capable language model.

```javascript
script({
    title: "explain-diagram",
    description: "Given an image of a diagram,
    explain what it contains",
    model: "vision",
})

defImages(env.files)

$``You are a helpful assistant. Your goal
is to look at the image provided and write
a description of what it contains. You
should infer the context of the diagram,
and write a thorough description of what
the diagram is illustrating.``
```

--------------------------------

### Full Source Code Example (MTS)

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/guides/images-in-azure-blob-storage.mdx

Provides the complete source code for the Azure Blob Storage image handling example in a GenAIScript module (.mts file).

```js
import { Code } from '@astrojs/starlight/components';
import { Steps } from "@astrojs/starlight/components"
import source from "../../../packages/sample/genaisrc/azure-blobs.genai.mts?raw"

// ... rest of the code as shown in the document ...
```

--------------------------------

### Run 'Say "hello!" in emojis' with GenAIScript

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/getting-started/tutorial.md

This snippet demonstrates the basic usage of the `$` function in GenAIScript to generate a user message that prompts the LLM to say 'hello!' in emojis. The output shows the generated user message and the LLM's response.

```javascript
$"Say \"hello!\" in emojis"

```

--------------------------------

### Example Markdown with Frontmatter

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/guides/detection-outdated-descriptions.md

This is an example of a markdown file structure that includes a frontmatter header. The frontmatter contains metadata such as title and description, which are often used by documentation systems.

```markdown
---
title: "My Document"
description: "This is a sample document."
---

# My Document

Lorem ipsum dolor sit amet, consectetur adipiscing elit.
```

--------------------------------

### Development Environment and Execution

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/glossary.md

Covers the Playground, remote repository execution, and the 'run' function.

```APIDOC
## Development Environment and Execution

### Description
Details on the self-hosted Playground, running scripts from remote repositories, and the 'run' function for script execution.

### Features
*   **Playground**: Self-hosted web app for running GenAIScript scripts with a user-friendly UI.
*   **Remote repository**: Playground runs scripts from a remote repo using secrets from `.env`.
*   **run**: The `run` function wraps the [CLI run](/genaiscript/reference/cli/run) command to execute scripts.
*   **Run a script**: Executes a script, streaming LLM output to stdout from the workspace root.
*   **Running scripts from a remote repository**: Use `--remote` to load/run scripts from a remote repository via shallow clone.
```

--------------------------------

### Start a Persistent Container

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/reference/scripts/container.md

Starts a Docker container and prevents it from being automatically removed when it's no longer needed. The `persistent: true` option ensures the container remains available until manually stopped or removed.

```javascript
const container = await host.container({ persistent: true })
```

--------------------------------

### Identify Implementation Challenges with GenAIScript

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/blog/drafts/-understanding-and-debugging-genaiscript-prompt-chains-step-by-step-insights-into-multi-prompt-workflows-.md

Identifies potential implementation challenges by passing the 'designProposal' to another LLM prompt. This step showcases how to use outputs from previous stages to elicit critical analysis and identify risks within the workflow.

```javascript
// Step 3: Identify possible implementation challenges
const challenges = await $`You are a software architect. Review the design proposal below. List 2-3 potential implementation challenges or risks. Proposal: ${designProposal}`
```

--------------------------------

### Basic Azure OpenAI Script Setup

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/configuration/azure-openai.mdx

This snippet shows the basic syntax for initializing a script with an Azure OpenAI deployment. It assumes the necessary environment variables for Azure OpenAI are configured.

```javascript
script({ model: "azure:deployment-id" })

```

--------------------------------

### Example HTML Image Tag

Source: https://github.com/microsoft/genaiscript/blob/main/slides/pages/image-alt-text.md

An example of an HTML image tag with a basic, non-descriptive alt text attribute. This highlights the need for better alt text generation.

```html
<img src="..." alt="An image" />
```

--------------------------------

### QA Generator Example (JavaScript)

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/reference/scripts/chat-participants.mdx

An example script demonstrating the use of multi-turn chat for generating questions, answers, and validating answer quality. This snippet showcases a practical application of the `defChatParticipant` functionality.

```js
import { Code } from "@astrojs/starlight/components"
import scriptSource from "../../../../../../packages/sample/genaisrc/multi-turn.genai.mts?raw"

// ... (rest of the example code if available in scriptSource)
```

--------------------------------

### TLA+ Operator Definition Example

Source: https://github.com/microsoft/genaiscript/blob/main/slides/pages/tla-plus.md

This snippet shows a TLA+ operator definition for `passMsg`. It includes a prose comment explaining its implementability limitations related to atomic buffer operations in network interfaces.

```plaintext
  define {
    (*
      The passMsg operator is not implementable -at least not without using extra synchronization- because it atomically reads a message
      from the nic's in-buffer and writes to its out-buffer!
    *)
    passMsg(net, from, oldMsg, to, newMsg) == [ net EXCEPT ![from] = BagRemove(@, oldMsg), ![to] = BagAdd(@, newMsg) ]
  }
```

--------------------------------

### Prompt and API Examples for Poem Generation

Source: https://github.com/microsoft/genaiscript/blob/main/slides/pages/poem2.md

This snippet shows the text prompt used for poem generation and an example of the OpenAI API request and response. The response includes the content to be saved to 'poem.txt'.

```text
// prompt.txt
Write a poem using emojis.
Save it t file poem.txt.

```

```json
// OpenAI API request
{ "messages": [
    { "role": "system", "content": "Use this format for files..."} },
    { "role": "user", "content": "Write a poem using emojis. Save it to file poem.txt."}, 

```

```text
// OpenAI Text Response
File: ./poem.txt
🌅🌻🌞 🌳🍃🍂 🌙✨🌌 💤🌠🌙

```

--------------------------------

### Run GenAIScript Tests via CLI

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/getting-started/testing-scripts.mdx

Provides the command-line interface command to execute tests for a GenAIScript. This command initiates the testing process for a specified script.

```bash
npx genaiscript test proofreader
```

--------------------------------

### Run GenAIScript with custom parameters

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/guides/search-and-transform.mdx

This command-line example shows how to execute the GenAIScript 'st' script using the `--vars` option to provide specific values for the 'pattern' and 'transform' parameters. This allows for dynamic application of search and transformation logic.

```sh
genaiscript st --vars 'pattern=host\.exec\s*\([^,]+,\s*\[[^]]+\]\s*\)' 'transform=Convert the call to a single string command shell in TypeScript'
```

--------------------------------

### Get Node.js Project Information

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/reference/scripts/system.mdx

Retrieves and displays information about the current Node.js project, including Node.js version, npm version, package name, and package version. It executes shell commands to get this data.

```typescript
system({
    title: "Information about the current project",
})

export default async function (ctx: ChatGenerationContext) {
    const { $ } = ctx

    const { stdout: nodeVersion } = await host.exec("node", ["--version"])
    const { stdout: npmVersion } = await host.exec("npm", ["--version"])
    const { name, version } = (await workspace.readJSON("package.json")) || {}
    if (nodeVersion) $`- node.js v${nodeVersion}`
    if (npmVersion) $`- npm v${npmVersion}`
    if (name) $`- package ${name} v${version || ""}`
}
```

--------------------------------

### Create a 'hello world' poem prompt

Source: https://github.com/microsoft/genaiscript/blob/main/README.md

This snippet demonstrates the basic usage of the `$` template tag to create a prompt for an LLM. The prompt is sent to a configured LLM, which then generates the requested poem.

```javascript
`$"Write a 'hello world' poem."`
```

--------------------------------

### Install Playwright browsers and dependencies

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/reference/scripts/browser.md

This command installs the necessary browser binaries (like Chromium) and system dependencies required for Playwright to function. It's a crucial step for ensuring browser automation tools work correctly.

```bash
npx playwright install --with-deps chromium
```

--------------------------------

### Deploy Basic Linux Web App with Bicep

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/case-studies/bicep-best-practices.mdx

A Bicep file template for deploying a basic web application with a Linux app service plan on Azure. This is a standard sample template for Bicep.

```bicep
resource webApp 'Microsoft.Web/sites@2022-03-01' = {
  kind: 'app,linux'
  location: 'eastus'
  name: 'webapp-basic-linux'
  properties: {
    serverFarmId: asp.id
    siteConfig: {
      linuxFxVersion: 'DOTNET||6.0'
    }
  }
}

resource asp 'Microsoft.Web/serverfarms@2022-03-01' = {
  location: 'eastus'
  name: 'asp-basic-linux'
  sku: {
    name: 'F1'
    tier: 'Free'
  }
  kind: 'app'
  properties: {
    reserved: true
    linuxWorkersEnabled: true
  }
}
```

--------------------------------

### GenAIScript Configuration: Model Encodings Example

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/reference/configuration-files.mdx

This JSON configuration shows the `modelEncodings` property in GenAIScript, which allows you to specify custom encoding names for models. This example maps a specific Azure model name to a general encoding name.

```json
{
    "modelEncodings": {
        "azure:gpt__4o_random_name": "gpt-4o"
    }
}
```

--------------------------------

### Automate with GenAIScript API

Source: https://github.com/microsoft/genaiscript/blob/main/README.md

Automates GenAIScript tasks programmatically using the API. This example shows how to import the 'run' function and execute the 'tlaplus-linter' command.

```javascript
import { run } from "genaiscript/api"

const res = await run("tlaplus-linter", "*.tla")
```

--------------------------------

### Start a Container with Port Bindings

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/reference/scripts/container.md

Starts a Docker container with network access enabled and maps a specific port on the host machine to a port inside the container. This allows external access to services running within the container, such as web servers.

```javascript
const container = await host.container({
    networkEnabled: true,
    ports: {
        containerPort: "80/tcp",
        hostPort: 8088,
    }, // array also supported
})
```

--------------------------------

### TypeScript Entrypoint Definition with Checkpointer

Source: https://github.com/microsoft/genaiscript/blob/main/packages/sample/genaisrc/langgraph.md

Defines a GenAIScript workflow using the `entrypoint` function. This example shows how to initialize a workflow with a `MemorySaver` checkpointer for persistence and enables features like human-in-the-loop. The workflow function accepts a single input and must return JSON-serializable data.

```typescript
import { entrypoint, MemorySaver } from "@langchain/langgraph";

const checkpointer = new MemorySaver();

const myWorkflow = entrypoint(
  { checkpointer, name: "myWorkflow" },
  async (someInput: Record<string, any>): Promise<number> => {
    // some logic that may involve long-running tasks like API calls,
    // and may be interrupted for human-in-the-loop.
    return result;
  }
);

```

--------------------------------

### Add GenAIScript CLI to package.json Scripts

Source: https://github.com/microsoft/genaiscript/blob/main/docs/src/content/docs/reference/cli/index.mdx

Configures `package.json` to automatically run GenAIScript commands after installation or updates, and provides a shorthand for invoking the CLI. The `postinstall` and `postupdate` scripts ensure TypeScript definition files are fixed, while the `genaiscript` script allows running commands like `npm run genaiscript ...`.

```json
{
    "scripts": {
        "postinstall": "genaiscript scripts fix",
        "postupdate": "genaiscript scripts fix",
        "genaiscript": "genaiscript"
    }
}
```

--------------------------------

### Example of Marked Code for Debugging

Source: https://github.com/microsoft/genaiscript/blob/main/slides/slides-ast.md

Shows an example of how code is transformed during the debug statement instrumentation process. It includes the original code, the intermediate 'marked' state with unique identifiers, a sample LLM response, and the final updated code with descriptive debug statements.

```javascript
if (condition) {
  ...
}
```

```javascript
if (condition) {
  dbg(<DBG_ID_1>);
  ...
}
```

```text
DBG_ID_1: reason
```

```javascript
if (condition) {
  dbg("...describe why we took the branch...");
  ...
}
```

--------------------------------

### TypeScript Greeter Class Definition

Source: https://github.com/microsoft/genaiscript/blob/main/packages/sample/src/greeter.ts.slides.md

Defines a basic TypeScript class named 'Greeter' with a 'greeting' property. It includes a constructor to initialize this property and a 'greet' method to generate a greeting message. This is a foundational example for object-oriented programming in TypeScript.

```typescript
class Greeter {
    greeting: string;
    
    constructor(message: string) {
      this.greeting = message;
    }

    greet() {
      return "Hello, " + this.greeting;
    }
}
```