Actions & Replay

brow actions

View the action log — every navigate, click, fill, key, select, upload, and fetch recorded in this session.

brow actions -s <id> [--json] [--clear]

Flag	Description
--json	Output as JSON array instead of human-readable format
--clear	Reset the action log

brow actions -s 1
# 1   navigate  https://example.com  [200]
# 2   fill      #email  value='user@example.com'
# 3   fill      #password  value='...'
# 4   click     button[type=submit]
# 5   navigate  https://example.com/dashboard  [200]

brow actions -s 1 --json
# [
#   {"seq": 1, "action": "navigate", "url": "https://example.com", "status": 200},
#   {"seq": 2, "action": "fill", "selector": "#email", "value": "user@example.com"},
#   ...
# ]

The action log is the input for the Playbook Writer workflow. After an exploratory session, review the log to identify the minimal set of steps needed, then write a playbook YAML from them.

brow replay

Execute a playbook YAML file against the current session.

brow replay -s <id> <playbook.yaml> [--var <key>=<value>]

Argument/Flag	Description
<playbook.yaml>	Path to the playbook file
--var	Override a playbook variable (repeatable)

brow replay -s 1 search.yaml
brow replay -s 1 search.yaml --var query="playwright"
brow replay -s 1 search.yaml --var query="brow" --var count=20

Output per step:

✓  navigate   https://example.com  200
✓  fill       #search
✓  key        Enter
✓  navigate   https://example.com/results  200
   → {"results": [{"title": "brow", "url": "..."}...
✗  click      .nonexistent   Timeout waiting for selector

Playbook format

name: search-example
description: Search example.com for a term
base_url: https://example.com
auth: none                    # none | browser-session | browser
vars:
  query: "default value"      # overridable with --var

steps:
  - action: navigate
    url: /search

  - action: fill
    selector: "input[name=q]"
    value: "{query}"

  - action: key
    key: Enter

  - action: wait               # by condition, not just a fixed sleep
    selector: "#results"
    state: visible             # visible | hidden | attached | detached
    timeout: 10000

  - action: fetch
    url: /api/results?q={query}
    method: GET
    headers:
      X-Api-Key: "{api_key}"
    output: results            # captured JSON becomes {results} (and {results[key]}) in later steps

Supported step actions

Action	Required fields	Optional fields
navigate	url	timeout
click	selector	—
fill	selector, value	—
key	key	—
select	selector, value	—
fetch	url	method, headers, auth, output, expect_status
wait	—	selector+state, or ms for a fixed sleep
assert	selector	state — fails the step (not the run, unless stop_on_failure) if unmet
for_each	var, items, steps	loop nested steps over a literal or captured list

A top-level auth: none on the playbook applies to every fetch step that doesn’t set its own auth. Add stop_on_failure: true at the top level to halt the whole run (including out of a for_each) after the first failed step, instead of the default “record it and keep going.”

For anything needing real branching or a loop body more complex than a few flat steps, prefer brow run <file.py> — a Python file executed once against the live session, with the same page/context/browser/state/pages variables as eval, plus args from --arg key=value — over growing this YAML further.

Variable substitution

Use {varname} anywhere in string fields. Variables come from the vars section and can be overridden with --var:

vars:
  user_id: "123"
  date: "2024-01-01"

steps:
  - action: fetch
    url: /api/users/{user_id}/history?from={date}

brow replay -s 1 playbook.yaml --var user_id=456 --var date=2024-06-01