Multi Config¶

The multi command lets you execute multiple actions from a single YAML config file. All actions run sequentially on a single browser session, avoiding the overhead of launching a browser for each action.

Usage¶

wavexis multi <config.yml>

Config format¶

The YAML file must have an actions key containing a list of action entries. Each entry is a dict with a single key (the action type) and a dict of parameters.

actions:
  - screenshot:
      url: https://example.com
      full_page: true
  - pdf:
      url: https://example.com
      paper: a4
  - eval:
      url: https://example.com
      expression: document.title
  - dom:
      url: https://example.com
      action: get
      selector: h1
  - navigate:
      url: https://example.org
  - scrape:
      urls:
        - https://example.com
        - https://example.org
      expression: document.title

How it works¶

wavexis parses the YAML file and validates its structure.
A single browser instance is launched.
Each action is dispatched to the corresponding action class (ScreenshotAction, EvalAction, etc.).
Actions execute sequentially — each one completes before the next starts.
Results are collected and printed. Binary results (screenshots, PDFs) show byte counts; text results show a preview.
The browser is closed after all actions complete.

Watch mode¶

Re-execute the config automatically when the file changes. This is designed for iterative config development — edit the YAML, save, and see results immediately:

wavexis multi config.yml --watch

Output:

Watching config.yml for changes (Ctrl+C to stop)...
[14:32:05] Completed 3 actions
  Waiting for changes...
  File changed, re-running...
[14:32:18] Completed 3 actions
  Waiting for changes...

Watch mode uses file polling (1-second interval) for cross-platform compatibility. Press Ctrl+C to stop.

Dry run¶

Validate the config and show the planned actions without launching a browser:

wavexis multi config.yml --dry-run

Output:

Plan: 3 action(s)
  1. screenshot: https://example.com (full_page=True)
  2. eval: https://example.com — document.title
  3. pdf: https://example.com (paper=a4)

Dry run is useful for:

Validating config syntax — catches structural errors before running.
Reviewing action sequences — verify the order and parameters of actions.
CI validation — ensure configs are valid in pull requests without executing them.

Supported actions¶

Action	Parameters	Description
`screenshot`	`url`, `full_page`, `format`	Take a screenshot
`pdf`	`url`, `paper`, `landscape`	Generate a PDF
`eval`	`url`, `expression`, `await_promise`	Evaluate JavaScript
`dom`	`url`, `action`, `selector`	DOM operations
`navigate`	`url`	Navigate to a URL
`scrape`	`urls`, `expression`	Batch scrape multiple URLs
`click`	`url`, `selector`	Click an element
`type`	`url`, `selector`, `text`	Type text into an element
`cookies`	`url`, `action`, `cookie`, `name`, `domain`	Cookie operations
`headers`	`url`, `action`, `headers`, `user_agent`	HTTP headers and user agent
`wait`	`strategy`, `selector`, `url_pattern`, `timeout`	Wait for a condition
`emulation`	`action`, `device`, `width`, `height`, `timezone`, `dark_mode`, ...	Device/viewport/timezone/geolocation/dark mode emulation

Cookies in multi¶

The cookies action supports four operations: get, set, delete, clear.

Get cookies¶

actions:
  - cookies:
      url: https://example.com
      action: get

actions:
  - cookies:
      url: https://example.com
      action: set
      cookie:
        name: session_token
        value: abc123
        domain: .example.com
        path: /

actions:
  - cookies:
      url: https://example.com
      action: delete
      name: session_token
      domain: .example.com

Clear all cookies¶

actions:
  - cookies:
      url: https://example.com
      action: clear

Headers in multi¶

The headers action supports two operations: set-headers (set extra HTTP headers) and set-user-agent (override the User-Agent string).

Set extra HTTP headers¶

actions:
  - headers:
      url: https://example.com
      action: set-headers
      headers:
        X-Custom-Header: my-value
        Authorization: Bearer token123
  - screenshot:
      url: https://example.com

Headers are applied before subsequent actions in the sequence, so the screenshot will be taken with the custom headers active.

Override User-Agent¶

actions:
  - headers:
      url: https://example.com
      action: set-user-agent
      user_agent: "Mozilla/5.0 (iPhone; CPU iPhone OS 17_0 like Mac OS X)"
  - screenshot:
      url: https://example.com

Wait in multi¶

The wait action pauses execution until a condition is met. This is essential for pages with dynamic content that loads after navigation.

Wait strategies¶

Strategy	Description	Required params
`load`	Wait for page load event	none
`domcontentloaded`	Wait for DOM content loaded	none
`networkidle`	Wait until network is idle	none
`selector`	Wait for a CSS selector to appear	`selector`
`url`	Wait for URL to match a pattern	`url_pattern`

Examples¶

actions:
  - navigate:
      url: https://app.example.com
  - wait:
      strategy: selector
      selector: "#dashboard-loaded"
      timeout: 10000
  - screenshot:
      url: https://app.example.com
      full_page: true

actions:
  - click:
      url: https://app.example.com
      selector: "#load-more"
  - wait:
      strategy: networkidle
      timeout: 5000
  - scrape:
      urls:
        - https://app.example.com
      expression: "document.querySelectorAll('.item').length"

Emulation in multi¶

The emulation action sets device, viewport, geolocation, timezone, or dark mode before subsequent actions. This is useful for testing how a page renders under different conditions.

Device emulation¶

actions:
  - emulation:
      action: device
      device: iphone-15
  - screenshot:
      url: https://example.com
      full_page: true

Viewport override¶

actions:
  - emulation:
      action: viewport
      width: 1920
      height: 1080
  - screenshot:
      url: https://example.com

Timezone¶

actions:
  - emulation:
      action: timezone
      timezone: America/New_York
  - eval:
      url: https://example.com
      expression: new Date().getTimezoneOffset()

Dark mode¶

actions:
  - emulation:
      action: dark_mode
      dark_mode: true
  - screenshot:
      url: https://example.com

Geolocation¶

actions:
  - emulation:
      action: geolocation
      latitude: 37.7749
      longitude: -122.4194
      accuracy: 100
  - eval:
      url: https://example.com
      expression: navigator.geolocation

Variables¶

Configs support variable substitution using {{var}} syntax. Define variables in a top-level vars section and reference them in any action parameter.

User-defined variables¶

vars:
  target_url: https://app.example.com
  username: admin@example.com
  password: secret123

actions:
  - navigate:
      url: "{{target_url}}/login"
  - type:
      url: "{{target_url}}/login"
      selector: "#username"
      text: "{{username}}"
  - type:
      url: "{{target_url}}/login"
      selector: "#password"
      text: "{{password}}"
  - click:
      url: "{{target_url}}/login"
      selector: "#submit"
  - screenshot:
      url: "{{target_url}}/dashboard"
      full_page: true

Environment variables¶

Access environment variables with {{env.KEY}}:

actions:
  - eval:
      url: "{{env.TARGET_URL}}"
      expression: document.title

TARGET_URL=https://staging.example.com wavexis multi config.yml

Substitution rules¶

Variables are substituted in string values only — dicts, lists, and scalars are left as-is.
Unknown variables are left unchanged (e.g. {{unknown}} stays literal).
Missing environment variables are left unchanged.
Substitution is recursive — variables in nested dicts and lists are all processed.

Parallel execution¶

By default, actions execute sequentially. Use --parallel to run all actions concurrently on the same backend:

wavexis multi config.yml --parallel

This is useful when actions are independent (e.g. scraping multiple URLs). Results are returned in the same order as the actions in the config.

When to use parallel¶

Independent actions — scraping different URLs, taking screenshots of different pages.
Performance — reduce total execution time when actions don't depend on each other.

When NOT to use parallel¶

Sequential dependencies — if action B depends on action A's side effects (e.g. login then screenshot).
State mutations — cookies, headers, or emulation that affect subsequent actions.

Parallel with watch¶

wavexis multi config.yml --parallel --watch

Action caching¶

Cache action results to avoid re-analyzing pages when running workflows repeatedly. Useful in watch mode or CI where the same config is executed multiple times.

wavexis multi config.yml --cache-ttl 60

Cacheable actions¶

Action	Cached
`screenshot`	Yes
`dom`	Yes
`scrape`	Yes
`eval`	Yes
`cookies`	Yes
`headers`	Yes
`navigate`	No
`click`	No
`type`	No
`pdf`	No
`wait`	No
`emulation`	No

How caching works¶

Cache is keyed by URL, action type, and a hash of action parameters.
TTL (time-to-live) controls how long cached results are valid.
On cache hit, the action is skipped and the cached result is returned.
Cache is in-memory per multi invocation — no persistent storage.
Use --cache-ttl 0 to disable caching (default).

Example: iterative development¶

# Cache for 120 seconds while iterating on config
wavexis multi config.yml --watch --cache-ttl 120

Each time the config changes and re-runs, cacheable actions that haven't expired will return instantly.

Validation errors¶

Invalid config files raise MultiConfigError with exit code 2:

Missing actions key
actions is not a list
Action entry has multiple keys
Action parameters are not a dict
Unknown action type

A complete login flow with cookie inspection:

actions:
  - navigate:
      url: https://app.example.com/login
  - type:
      url: https://app.example.com/login
      selector: "#username"
      text: admin@example.com
  - type:
      url: https://app.example.com/login
      selector: "#password"
      text: secret123
  - click:
      url: https://app.example.com/login
      selector: "#login-button"
  - eval:
      url: https://app.example.com/dashboard
      expression: document.title
  - cookies:
      url: https://app.example.com/dashboard
      action: get
  - screenshot:
      url: https://app.example.com/dashboard
      full_page: true

wavexis multi login-flow.yml

Output:

Completed 7 actions
  Action 1: None
  Action 2: None
  Action 3: None
  Action 4: None
  Action 5: Dashboard - Example App
  Action 6: [{"name": "session", "value": "abc123", ...}]
  Action 7: 45678 bytes

Multi Config¶

Usage¶

Config format¶

How it works¶

Watch mode¶

Dry run¶

Supported actions¶

Cookies in multi¶

Get cookies¶

Set a cookie¶

Delete a cookie¶

Clear all cookies¶

Headers in multi¶

Set extra HTTP headers¶

Override User-Agent¶

Wait in multi¶

Wait strategies¶

Examples¶

Emulation in multi¶

Device emulation¶

Viewport override¶

Timezone¶

Dark mode¶

Geolocation¶

Variables¶

User-defined variables¶

Environment variables¶

Substitution rules¶

Parallel execution¶

When to use parallel¶

When NOT to use parallel¶

Parallel with watch¶

Action caching¶

Cacheable actions¶

How caching works¶

Example: iterative development¶

Validation errors¶

Example: login flow¶