Introduction | Basic tool display | Tool title and icon | Via tool annotations | Present tense while running, past tense once done | Via the tool result | Tool intent | Customizing tool result display | Activity-row label and value preview | Rich content in the drill-down card | Alternative HTML display | Alternative markdown display | Fully custom standalone output | Display Options | Grouping tool calls | Global display options

Start with a chat you set up by hand | Save the setup in btw.md | Create and edit the file | Choose the model with client | Choose tools with tools | Set btw options with options | Write instructions in the body | Where btw looks for btw.md | Next: the same format for agents

Prerequisites | Create the analysis script | Run the analysis script

Design philosophy | Working with the agent skill | Trust flow | Information layers | Semantic layer | Context layer | Examples | Data dictionary | Measure file | Context document | Warehouse semantic layers | Data sources | Data dictionaries | Definitions | Project directory organization | Constructing the agent

commons' harness | SQL code execution | Destructive actions | Data access | Transmitting results to the R process | R code execution | OS-level sandboxing | Communication with the main app process | Permissioning facts | Logging trajectories

An example: reviewing a report | Give the main chat access to agents | Create an agent file | Add the other two agents | Give an agent its own model | Run the review | Work with an agent directly | Recap

A minimal app | Inputs | Outputs | Messages | Traditional Shiny renderers | Bookmarking | Next steps

Annotations | Annotation settings | Use roxygen prefix | Annotations example | Equivalent programmatic usage | Specifying route name | Specifying route order | Specifying root url for the whole file | Global annotations | Handler annotations | Endpoint | More details on Type | Asset annotation | Equivalent programmatic usage (note that argument order is reversed) | Auth guard annotation | Datastore annotation | plumber2 annotation | Forwards and redirects | Annotation example | Shiny | Reports

Creating and Controlling routers | Defining handlers | Listening for and triggering events | Static File Routers

The Response Object | Serializers | Standard serializers | Boxed vs Unboxed JSON | Graphics serializers | Bypassing Serialization | Custom serializers | Error Handling | Setting Cookies | Setting Unencrypted Cookies | Setting Encrypted Cookies | Documenting responses

Handlers | Handler methods | Paths | Path parameters | Path wildcard | Path priority | Handler input | Handler arguments | query | body | request | Cookies | Headers | response | server | client_id | Type casting input | Defaults | Required parameters | Static File Handler

Starter template | Chat-first page | Deferred data sources | Reactives | Filtered data | SQL query | Title | Custom UI | Data views | Programmatic updates | Multiple tables | Registering tables | Per-table reactive access | Tracking the active table | Data dictionary | Separate chat interfaces | See also

Setup | Choose a model | Set up credentials | Create a basic chatbot | Inspect the code | Add a system prompt | Add greetings and suggestions | On startup | Mid-conversation | Layouts | Full-window page with navigation | Screen-filling layout | Sidebar layout | Card layout | Slash commands | Prompt expansion | Side effects | Client-side handlers | Key points | Stream cancellation | Using chat_server() (recommended) | Manual approach | File attachments

A first example | Comments | Collections | Sequences: YAML’s ordered collections | Mappings: key/value pairs | Scalars | Block scalars | Quoted scalars | Plain (unquoted) scalars | End-to-end example | Why YAML 1.2? | YAML 1.1 versus 1.2 quick reference | Quick notes

Tags in YAML and how yaml12 handles them | Using handlers to transform tagged nodes while parsing | Post-process tags yourself | Mappings revisited: non-string keys and yaml_keys | Tagged mapping keys | Document Streams and Markers | Reading Multiple Documents | Writing Multiple Documents | Writing YAML with tags | Anchors | Debugging | (Very) Advanced Tags | Tag directives (%TAG) | TAG URIs | Core schema tags

Specify a model | Credentials | Recommended models

The basics | Posit Connect | Multiple clients and R sessions | Custom tools

Provide a greeting | Generate a greeting | Greetings with multiple tables

Data dictionary | Format | Tables | Relationships | Glossary | Data description | Additional instructions | Default prompt | Custom template

Schema retrieval | Data updating | Data analysis | Data visualization | View the source

Data frames | Database connections | DuckDB | SQLite | PostgreSQL | MySQL | Creating a database from a data frame | Pins | Lazy queries with pins | Custom sources

Getting started | Choosing tools | Custom apps | What you can ask for | Chart controls | How it works | See also

Registering the engine | The {ggsql} chunk | Chunk options | Defining a backend | Choosing a writer | Communicating with R and Python | Getting data back from ggsql

Package anatomy | Creating a reader | Executing a query | Rendering a plot

The _server.yml file | Creating a _server.yml file | Testing your _server.yml file | Posit Connect | DigitalOcean | Prerequisites | Basic Deployment | Learn More | Docker | Choosing a Base Image | System Dependencies | Example Dockerfile | pm2 | Setting up pm2 for plumber2 | Starting Your API with pm2 | Managing Your API | Running Multiple Instances | Automatic Startup on Boot | Using an Ecosystem File | Monitoring and Logs | systemd | Creating a systemd Service | Setting Up Your Application | Managing the Service | Viewing Logs | Service Management Commands

What does it mean to extend plumber2 | Adding new tags | Tag registration | New block type | Existing block type | apply_plumber2_block methods | Single main class | Subclass | Adding a programmatic interface | New parsers, serializers, and async engines | Serializers | Graphics output | Parsers | Asynchronous engines | The Plumber2 object and its extension points

Traces | Subtraces | Metrics | Logs

Overview | Creating a brand | ggplot2 | thematic | ggiraph | plotly | gt | flextable

Specifying the Inputs | Customizing The Output

Networking & Firewalls | HTTPS | Denial Of Service (DoS) | Sanitization & Injection | Security Headers | Cross-Origin Resource Sharing (CORS) | Resource Isolation Policies | Cookies

The standard | The _server.yml file | The engine | The server | Recommendations for _server.yml engines | File paths | Options | Case study: plumber2 | YAML content | launch_server() | Supported frameworks

Execution Model | Environments | Performance & Request Processing | Managing State | In-Memory | File System | Cookies | External Data Store | Exit Handlers

Web APIs

Debugging | Print Debugging | Router Stage Debugging | Interactive Debugging | Port Range

Moving from plumber | pr_*() to api_*() | Only path arguments are passed as named arguments to your handler | No more filter and preempt | forward replaced by Next and Break | New Request and Response objects | New Plumber object | Added @query and @body tags

Connecting | Content Management

Getting Started | Custom CA Bundle | Turn Off Certificate Trust Verification - ONLY FOR TESTING | Custom Headers, Cookies, Proxy, etc. | Using Kerberos