nodriver-browser

Name: nodriver-browser
Author: appautomaton

5installs1forks

How to install nodriver-browser?

appautomaton/webmaton/nodriver-browser

$npx -y skills add appautomaton/webmaton --skill nodriver-browser

Installs into the current project.

Summary

Persistent Chrome/Chromium browser automation skill built on nodriver. Use when a page needs JavaScript rendering, authorized login/session continuity, clicking or typing, DOM snapshots with stable refs, screenshots, or multi-step look-think-act flows that ordinary WebFetch/searc

Files · 1

View on GitHub

SKILL.md

1	# nodriver-browser
2
3	A persistent Chrome/Chromium browser that stays alive between Claude's turns. Built on `nodriver` (CDP-direct, no Selenium, no `navigator.webdriver`). Every script attaches to the same long-running browser, performs one action, exits — the browser and its tab keep going.
4
5	Core invariant: ONE daemon, ONE persistent tab (`tabs[0]`). Every script reports `tabs_open` in its output. If you ever see `tabs_open > 1`, treat it as a real signal that something opened a stray tab — read the warning and act on it.
6
7	## When to use this skill
8
9	- The page needs JavaScript to render (SPA, infinite scroll, lazy load)
10	- WebFetch is blocked by anti-bot systems (Cloudflare, DataDome, Imperva, hCaptcha)
11	- The task requires interaction: clicking buttons, filling forms, multi-step flows, dropdown selection
12	- You need session state across multiple actions (logged-in scraping, multi-page checkout, OAuth flows)
13	- You need visual proof of a page (screenshot) for debugging or reporting
14
15	## When NOT to use this skill
16
17	- Static HTML that loads fully on first GET → use `WebFetch`
18	- One-shot search query → use `WebSearch` (or grok-search skill if it exists)
19	- A JSON API endpoint → use `curl` via Bash, you don't need a browser at all
20	- A single quick scrape with no interactivity → consider a one-off Python script, not this skill
21
22	This skill spawns a Chromium process (~150-200 MB RAM) that stays alive until you explicitly stop it. Worth it for interactive flows; overkill for one URL.
23
24	## Quick start
25
26	Scripts live in `scripts/` next to this SKILL.md — resolve paths relative to the skill root.
27
28	Run them directly as executables; the `#!/usr/bin/env -S uv run --script`
29	shebang invokes uv and reads the PEP 723 metadata for you. Do not run these
30	scripts with `python`, `python3`, or `python -m`; that bypasses the shebang,
31	dependency metadata, and Python version pin. If executable dispatch is
32	unavailable, use `uv run --script scripts/nav.py ...`.
33
34	```bash
35	# 1. Navigate (auto-starts daemon on first call — no manual start needed)
36	scripts/nav.py https://news.ycombinator.com
37
38	# Optional: visible browser window, using the user's Chrome profile
39	scripts/nav.py --headed --user-profile https://example.com
40
41	# 2. Snapshot the page — gives you text + numbered refs for every interactive element
42	scripts/snapshot.py
43
44	# 3. Click something by ref id from the snapshot
45	scripts/click.py r17
46
47	# 4. Read state any time — it's the same tab as before, even from a fresh process
48	scripts/state.py
49
50	# 5. When done with the session, stop the daemon (frees ~190 MB)
51	scripts/stop_daemon.py
52	```
53
54	## Script reference
55
56	All runnable scripts use `uv run --script` with PEP 723 metadata and return JSON to stdout. Invoke them directly, not through `python` or `python3`. Every script (except daemon control) appends `tabs_open: N` and emits a `warning` field if `N > 1`.
57
58	Leading browser options work on `start_daemon.py` and every script that auto-starts/attaches:
59
60	\| Option \| Purpose \|
61	\|---\|---\|
62	\| `--headless` \| Start a headless daemon. This is the default when no daemon is running. \|
63	\| `--headed` \| Start a visible Chrome/Chromium window. If a daemon is already running in headless mode, stop it first. \|
64	\| `--skill-profile` \| Use the isolated profile at `~/.cache/nodriver-skill/profile/`. This is the default. \|
65	\| `--user-profile` \| Use the user's Chrome profile root. Useful for existing logged-in state; requires that regular Chrome is not already locking the same profile. \|
66	\| `--profile-directory NAME` \| Use a Chrome profile directory such as `Default` or `Profile 1`; implies `--user-profile`. \|
67	\| `--user-data-dir PATH` \| Override the Chrome user-data root; implies `--user-profile`. \|
68	\| `--no-sandbox` \| Disable Chrome's OS sandbox. Use only when Chrome cannot start in constrained environments such as PRoot/container/root setups. Do not use for normal system Chrome or the user's Chrome profile. \|
69
70	Environment equivalents: `NODRIVER_SKILL_MODE=headed\|headless`, `NODRIVER_SKILL_PROFILE=skill\|user`, `NODRIVER_CHROME_PROFILE_DIRECTORY="Profile 1"`, `NODRIVER_CHROME_USER_DATA_DIR=/path/to/User Data`, `NODRIVER_CHROME_NO_SANDBOX=1`.
71
72	### Daemon control
73
74	\| Script \| Purpose \| Output \|
75	\|---\|---\|---\|
76	\| `start_daemon.py` \| Idempotent start. No-op if already running. Supports leading browser options. \| `{ok, pid, port, mode, profile, no_sandbox, already_running}` \|
77	\| `stop_daemon.py` \| K

People who liked this also viewed

6 picks

Type

nodriver-browser

5installs1forks

View on GitHub

How to install nodriver-browser?

appautomaton/webmaton/nodriver-browser

$npx -y skills add appautomaton/webmaton --skill nodriver-browser

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# nodriver-browser
2
3	A persistent Chrome/Chromium browser that stays alive between Claude's turns. Built on `nodriver` (CDP-direct, no Selenium, no `navigator.webdriver`). Every script attaches to the same long-running browser, performs one action, exits — the browser and its tab keep going.
4
5	Core invariant: ONE daemon, ONE persistent tab (`tabs[0]`). Every script reports `tabs_open` in its output. If you ever see `tabs_open > 1`, treat it as a real signal that something opened a stray tab — read the warning and act on it.
6
7	## When to use this skill
8
9	- The page needs JavaScript to render (SPA, infinite scroll, lazy load)
10	- WebFetch is blocked by anti-bot systems (Cloudflare, DataDome, Imperva, hCaptcha)
11	- The task requires interaction: clicking buttons, filling forms, multi-step flows, dropdown selection
12	- You need session state across multiple actions (logged-in scraping, multi-page checkout, OAuth flows)
13	- You need visual proof of a page (screenshot) for debugging or reporting
14
15	## When NOT to use this skill
16
17	- Static HTML that loads fully on first GET → use `WebFetch`
18	- One-shot search query → use `WebSearch` (or grok-search skill if it exists)
19	- A JSON API endpoint → use `curl` via Bash, you don't need a browser at all
20	- A single quick scrape with no interactivity → consider a one-off Python script, not this skill
21
22	This skill spawns a Chromium process (~150-200 MB RAM) that stays alive until you explicitly stop it. Worth it for interactive flows; overkill for one URL.
23
24	## Quick start
25
26	Scripts live in `scripts/` next to this SKILL.md — resolve paths relative to the skill root.
27
28	Run them directly as executables; the `#!/usr/bin/env -S uv run --script`
29	shebang invokes uv and reads the PEP 723 metadata for you. Do not run these
30	scripts with `python`, `python3`, or `python -m`; that bypasses the shebang,
31	dependency metadata, and Python version pin. If executable dispatch is
32	unavailable, use `uv run --script scripts/nav.py ...`.
33
34	```bash
35	# 1. Navigate (auto-starts daemon on first call — no manual start needed)
36	scripts/nav.py https://news.ycombinator.com
37
38	# Optional: visible browser window, using the user's Chrome profile
39	scripts/nav.py --headed --user-profile https://example.com
40
41	# 2. Snapshot the page — gives you text + numbered refs for every interactive element
42	scripts/snapshot.py
43
44	# 3. Click something by ref id from the snapshot
45	scripts/click.py r17
46
47	# 4. Read state any time — it's the same tab as before, even from a fresh process
48	scripts/state.py
49
50	# 5. When done with the session, stop the daemon (frees ~190 MB)
51	scripts/stop_daemon.py
52	```
53
54	## Script reference
55
56	All runnable scripts use `uv run --script` with PEP 723 metadata and return JSON to stdout. Invoke them directly, not through `python` or `python3`. Every script (except daemon control) appends `tabs_open: N` and emits a `warning` field if `N > 1`.
57
58	Leading browser options work on `start_daemon.py` and every script that auto-starts/attaches:
59
60	\| Option \| Purpose \|
61	\|---\|---\|
62	\| `--headless` \| Start a headless daemon. This is the default when no daemon is running. \|
63	\| `--headed` \| Start a visible Chrome/Chromium window. If a daemon is already running in headless mode, stop it first. \|
64	\| `--skill-profile` \| Use the isolated profile at `~/.cache/nodriver-skill/profile/`. This is the default. \|
65	\| `--user-profile` \| Use the user's Chrome profile root. Useful for existing logged-in state; requires that regular Chrome is not already locking the same profile. \|
66	\| `--profile-directory NAME` \| Use a Chrome profile directory such as `Default` or `Profile 1`; implies `--user-profile`. \|
67	\| `--user-data-dir PATH` \| Override the Chrome user-data root; implies `--user-profile`. \|
68	\| `--no-sandbox` \| Disable Chrome's OS sandbox. Use only when Chrome cannot start in constrained environments such as PRoot/container/root setups. Do not use for normal system Chrome or the user's Chrome profile. \|
69
70	Environment equivalents: `NODRIVER_SKILL_MODE=headed\|headless`, `NODRIVER_SKILL_PROFILE=skill\|user`, `NODRIVER_CHROME_PROFILE_DIRECTORY="Profile 1"`, `NODRIVER_CHROME_USER_DATA_DIR=/path/to/User Data`, `NODRIVER_CHROME_NO_SANDBOX=1`.
71
72	### Daemon control
73
74	\| Script \| Purpose \| Output \|
75	\|---\|---\|---\|
76	\| `start_daemon.py` \| Idempotent start. No-op if already running. Supports leading browser options. \| `{ok, pid, port, mode, profile, no_sandbox, already_running}` \|
77	\| `stop_daemon.py` \| K

People who liked this also viewed

6 picks

Type