397b0bcc9d
Agent bot conversations now feel more natural because AgentBot tokens can toggle typing status, so end users see a live typing indicator in the widget while the bot is preparing a reply. This keeps the interaction responsive and human-like without weakening token authorization boundaries. ## Closes - https://github.com/chatwoot/chatwoot/issues/8928 - https://linear.app/chatwoot/issue/CW-5205 ## How to test 1. Open the widget and start a conversation as a customer. 2. Connect an AgentBot to the same inbox. 3. Trigger `toggle_typing_status` with the AgentBot token (`typing_status: on`). 4. Confirm the customer sees the typing indicator in the widget. 5. Trigger `toggle_typing_status` with `typing_status: off` and confirm the indicator disappears. ## What changed - Added `toggle_typing_status` to bot-accessible conversation endpoints. - Restricted bot-accessible endpoint usage to `AgentBot` token owners only (non-user tokens like `PlatformApp` remain unauthorized). - Updated typing status flow to preserve AgentBot identity in dispatch/broadcast paths. - Added request coverage for AgentBot success and PlatformApp unauthorized behavior. - Added Swagger documentation for `POST /api/v1/accounts/{account_id}/conversations/{conversation_id}/toggle_typing_status` and regenerated swagger artifacts.
114 lines
6.5 KiB
Markdown
114 lines
6.5 KiB
Markdown
# Chatwoot Development Guidelines
|
||
|
||
## Build / Test / Lint
|
||
|
||
- **Setup**: `bundle install && pnpm install`
|
||
- **Run Dev**: `pnpm dev` or `overmind start -f ./Procfile.dev`
|
||
- **Seed Local Test Data**: `bundle exec rails db:seed` (quickly populates minimal data for standard feature verification)
|
||
- **Seed Search Test Data**: `bundle exec rails search:setup_test_data` (bulk fixture generation for search/performance/manual load scenarios)
|
||
- **Seed Account Sample Data (richer test data)**: `Seeders::AccountSeeder` is available as an internal utility and is exposed through Super Admin `Accounts#seed`, but can be used directly in dev workflows too:
|
||
- UI path: Super Admin → Accounts → Seed (enqueues `Internal::SeedAccountJob`).
|
||
- CLI path: `bundle exec rails runner "Internal::SeedAccountJob.perform_now(Account.find(<id>))"` (or call `Seeders::AccountSeeder.new(account: Account.find(<id>)).perform!` directly).
|
||
- **Lint JS/Vue**: `pnpm eslint` / `pnpm eslint:fix`
|
||
- **Lint Ruby**: `bundle exec rubocop -a`
|
||
- **Test JS**: `pnpm test` or `pnpm test:watch`
|
||
- **Test Ruby**: `bundle exec rspec spec/path/to/file_spec.rb`
|
||
- **Single Test**: `bundle exec rspec spec/path/to/file_spec.rb:LINE_NUMBER`
|
||
- **Run Project**: `overmind start -f Procfile.dev`
|
||
- **Ruby Version**: Manage Ruby via `rbenv` and install the version listed in `.ruby-version` (e.g., `rbenv install $(cat .ruby-version)`)
|
||
- **rbenv setup**: Before running any `bundle` or `rspec` commands, init rbenv in your shell (`eval "$(rbenv init -)"`) so the correct Ruby/Bundler versions are used
|
||
- Always prefer `bundle exec` for Ruby CLI tasks (rspec, rake, rubocop, etc.)
|
||
|
||
## Code Style
|
||
|
||
- **Ruby**: Follow RuboCop rules (150 character max line length)
|
||
- **Vue/JS**: Use ESLint (Airbnb base + Vue 3 recommended)
|
||
- **Vue Components**: Use PascalCase
|
||
- **Events**: Use camelCase
|
||
- **I18n**: No bare strings in templates; use i18n
|
||
- **Error Handling**: Use custom exceptions (`lib/custom_exceptions/`)
|
||
- **Models**: Validate presence/uniqueness, add proper indexes
|
||
- **Type Safety**: Use PropTypes in Vue, strong params in Rails
|
||
- **Naming**: Use clear, descriptive names with consistent casing
|
||
- **Vue API**: Always use Composition API with `<script setup>` at the top
|
||
|
||
## Styling
|
||
|
||
- **Tailwind Only**:
|
||
- Do not write custom CSS
|
||
- Do not use scoped CSS
|
||
- Do not use inline styles
|
||
- Always use Tailwind utility classes
|
||
- **Colors**: Refer to `tailwind.config.js` for color definitions
|
||
|
||
## General Guidelines
|
||
|
||
- MVP focus: Least code change, happy-path only
|
||
- No unnecessary defensive programming
|
||
- Ship the happy path first: limit guards/fallbacks to what production has proven necessary, then iterate
|
||
- Prefer minimal, readable code over elaborate abstractions; clarity beats cleverness
|
||
- Break down complex tasks into small, testable units
|
||
- Iterate after confirmation
|
||
- Avoid writing specs unless explicitly asked
|
||
- Remove dead/unreachable/unused code
|
||
- Don’t write multiple versions or backups for the same logic — pick the best approach and implement it
|
||
- Prefer `with_modified_env` (from spec helpers) over stubbing `ENV` directly in specs
|
||
- Specs in parallel/reloading environments: prefer comparing `error.class.name` over constant class equality when asserting raised errors
|
||
|
||
## Codex Worktree Workflow
|
||
|
||
- Use a separate git worktree + branch per task to keep changes isolated.
|
||
- Keep Codex-specific local setup under `.codex/` and use `Procfile.worktree` for worktree process orchestration.
|
||
- The setup workflow in `.codex/environments/environment.toml` should dynamically generate per-worktree DB/port values (Rails, Vite, Redis DB index) to avoid collisions.
|
||
- Start each worktree with its own Overmind socket/title so multiple instances can run at the same time.
|
||
|
||
## Commit Messages
|
||
|
||
- Prefer Conventional Commits: `type(scope): subject` (scope optional)
|
||
- Example: `feat(auth): add user authentication`
|
||
- Don't reference Claude in commit messages
|
||
|
||
## PR Description Format
|
||
|
||
- Start with a short, user-facing paragraph describing the product change.
|
||
- Add a `Closes` section with relevant issue links (GitHub, Linear, etc.).
|
||
- For feature PRs, add `How to test` from a product/UX standpoint.
|
||
- For bugfix PRs, use `How to reproduce` when helpful.
|
||
- Optionally add a `What changed` section for implementation highlights.
|
||
- Do not add a `How this was tested` section listing specs/commands.
|
||
|
||
## Project-Specific
|
||
|
||
- **Translations**:
|
||
- Only update `en.yml` and `en.json`
|
||
- Other languages are handled by the community
|
||
- Backend i18n → `en.yml`, Frontend i18n → `en.json`
|
||
- **Frontend**:
|
||
- Use `components-next/` for message bubbles (the rest is being deprecated)
|
||
|
||
## Ruby Best Practices
|
||
|
||
- Use compact `module/class` definitions; avoid nested styles
|
||
|
||
## Enterprise Edition Notes
|
||
|
||
- Chatwoot has an Enterprise overlay under `enterprise/` that extends/overrides OSS code.
|
||
- When you add or modify core functionality, always check for corresponding files in `enterprise/` and keep behavior compatible.
|
||
- Follow the Enterprise development practices documented here:
|
||
- https://chatwoot.help/hc/handbook/articles/developing-enterprise-edition-features-38
|
||
|
||
Practical checklist for any change impacting core logic or public APIs
|
||
- Search for related files in both trees before editing (e.g., `rg -n "FooService|ControllerName|ModelName" app enterprise`).
|
||
- If adding new endpoints, services, or models, consider whether Enterprise needs:
|
||
- An override (e.g., `enterprise/app/...`), or
|
||
- An extension point (e.g., `prepend_mod_with`, hooks, configuration) to avoid hard forks.
|
||
- Avoid hardcoding instance- or plan-specific behavior in OSS; prefer configuration, feature flags, or extension points consumed by Enterprise.
|
||
- Keep request/response contracts stable across OSS and Enterprise; update both sets of routes/controllers when introducing new APIs.
|
||
- When renaming/moving shared code, mirror the change in `enterprise/` to prevent drift.
|
||
- Tests: Add Enterprise-specific specs under `spec/enterprise`, mirroring OSS spec layout where applicable.
|
||
- When modifying existing OSS features for Enterprise-only behavior, add an Enterprise module (via `prepend_mod_with`/`include_mod_with`) instead of editing OSS files directly—especially for policies, controllers, and services. For Enterprise-exclusive features, place code directly under `enterprise/`.
|
||
|
||
## Branding / White-labeling note
|
||
|
||
- For user-facing strings that currently contain "Chatwoot" but should adapt to branded/self-hosted installs, prefer applying `replaceInstallationName` from `shared/composables/useBranding` in the UI layer (for example tooltip and suggestion labels) instead of adding hardcoded brand-specific copy.
|