Skip to content

Contributing to Canvas Chat

Plugin development

Plugins are the recommended way to extend Canvas-Chat functionality.

Canvas-Chat supports three levels of extensibility:

  1. Custom Node Types (Level 1): Add new visual node types with custom rendering and interactions
  2. Feature Plugins (Level 2): Create complex workflows, slash commands, and multi-step LLM interactions
  3. Extension Hooks (Level 3): Hook into existing features to modify or extend behaviors

Getting started with plugins

Start with these guides:

Plugin APIs

Complete API documentation:

Sharing plugins

To share your plugin with the community:

  1. Create a GitHub repository for your plugin
  2. Include clear README with installation and usage instructions
  3. Add example code and screenshots
  4. Open an issue in the main Canvas-Chat repo to announce your plugin
  5. Consider publishing to npm for easy installation

We maintain a list of community plugins in the documentation.

Code contributions

We are not accepting direct code contributions at this time.

Instead, we welcome detailed issue descriptions for new feature requests. The recommended workflow for proposing new features is:

  1. Clone the repository and explore the codebase
  2. Open an agentic coding harness (OpenCode, Claude Code, GitHub Copilot, Cursor, Windsurf, Antigravity, etc.)
  3. Use Plan Mode with a high-quality model (e.g., as of writing: Opus 4.5, GPT-5.2 (thinking), Gemini 3 Pro)
  4. Instruct the model to explore your feature request and ask clarifying questions
  5. Once the model has no more questions, have it post a detailed issue to the GitHub issue tracker with:
    • Clear description of the feature
    • Use cases and benefits
    • Technical considerations
    • Implementation approach (if explored)

This workflow ensures feature requests are well-thought-out and technically feasible before implementation.

Documentation Site

This directory contains the Canvas Chat documentation built with MkDocs and Material for MkDocs.

Structure

The documentation follows the Diataxis framework:

  • how-to/: Task-oriented guides for accomplishing specific goals
  • explanation/: Design decisions, architecture rationale, "why" documents
  • reference/: Technical descriptions of APIs, configuration options, data structures
  • releases/: Release notes and changelogs

Working with the docs

Prerequisites

Install the docs environment:

pixi install

Local development

Serve the docs locally with live reload:

pixi run -e docs docs-serve

Then open your browser to http://127.0.0.1:8000

Build the site

Build static HTML files:

pixi run -e docs docs-build

The built site will be in the site/ directory.

Deploy to GitHub Pages

Deploy to GitHub Pages (requires push access):

pixi run -e docs docs-deploy

This builds the site and pushes to the gh-pages branch.

Adding new pages

  1. Create a new .md file in the appropriate directory (how-to/, explanation/, reference/)
  2. Add the page to the nav section in mkdocs.yml
  3. Test locally with pixi run -e docs docs-serve

Configuration

The main configuration file is mkdocs.yml in the project root. It defines:

  • Site metadata (name, URL, repo)
  • Theme settings (Material theme with dark/light mode)
  • Navigation structure
  • Markdown extensions (admonitions, code highlighting, etc.)
  • Plugins (search)

Markdown extensions

The site supports:

  • Admonitions: !!! note, !!! warning, etc.
  • Code blocks with syntax highlighting
  • Tabbed content: === "Tab 1"
  • Code annotations: numbered callouts in code blocks
  • Table of contents: auto-generated from headers

See the Material for MkDocs documentation for details.