zerodom 0.0.1__tar.gz

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
@@ -0,0 +1,13 @@
1
+ __pycache__/
2
+ *.py[cod]
3
+ .venv/
4
+ .pytest_cache/
5
+ dist/
6
+ *.egg-info/
7
+
8
+ # session transcripts and scratch captures — local only
9
+ *-this-session-is-being-continued-from-a-previous-c.txt
10
+ assets/screenshot.png
11
+ # regenerable: `zerodom inspect --render --screenshot ... --html ... demo/demo-page.html`
12
+ demo/report.html
13
+ demo/annotated.png
zerodom-0.0.1/LICENSE ADDED
@@ -0,0 +1,95 @@
1
+ Business Source License 1.1
2
+
3
+ Parameters
4
+
5
+ Licensor: Syed Husnain Khalid
6
+
7
+ Licensed Work: ZeroDOM
8
+ The Licensed Work is (c) 2026 Syed Husnain Khalid
9
+
10
+ Additional Use Grant: You may make production use of the Licensed Work,
11
+ provided you do not offer the Licensed Work to third
12
+ parties as a hosted, managed, or embedded service where
13
+ the primary value of that service is the functionality
14
+ of the Licensed Work.
15
+
16
+ Change Date: 2030-08-09
17
+
18
+ Change License: Apache License, Version 2.0
19
+
20
+ For information about alternative licensing arrangements for the Licensed Work,
21
+ please contact the Licensor at contact@vexralabs.com.
22
+
23
+ -----------------------------------------------------------------------------
24
+
25
+ Business Source License 1.1
26
+
27
+ License text copyright © 2017 MariaDB Corporation Ab, All Rights Reserved.
28
+ "Business Source License" is a trademark of MariaDB Corporation Ab.
29
+
30
+ Terms
31
+
32
+ The Licensor hereby grants you the right to copy, modify, create derivative
33
+ works, redistribute, and make non-production use of the Licensed Work. The
34
+ Licensor may make an Additional Use Grant, above, permitting limited
35
+ production use.
36
+
37
+ Effective on the Change Date, or the fourth anniversary of the first publicly
38
+ available distribution of a specific version of the Licensed Work under this
39
+ License, whichever comes first, the Licensor hereby grants you rights under
40
+ the terms of the Change License, and the rights granted in the paragraph
41
+ above terminate.
42
+
43
+ If your use of the Licensed Work does not comply with the requirements
44
+ currently in effect as described in this License, you must purchase a
45
+ commercial license from the Licensor, its affiliated entities, or authorized
46
+ resellers, or you must refrain from using the Licensed Work.
47
+
48
+ All copies of the original and modified Licensed Work, and derivative works
49
+ of the Licensed Work, are subject to this License. This License applies
50
+ separately for each version of the Licensed Work and the Change Date may vary
51
+ for each version of the Licensed Work released by Licensor.
52
+
53
+ You must conspicuously display this License on each original or modified copy
54
+ of the Licensed Work. If you receive the Licensed Work in original or
55
+ modified form from a third party, the terms and conditions set forth in this
56
+ License apply to your use of that work.
57
+
58
+ Any use of the Licensed Work in violation of this License will automatically
59
+ terminate your rights under this License for the current and all other
60
+ versions of the Licensed Work.
61
+
62
+ This License does not grant you any right in any trademark or logo of
63
+ Licensor or its affiliates (provided that you may use a trademark or logo of
64
+ Licensor as expressly required by this License).
65
+
66
+ TO THE EXTENT PERMITTED BY APPLICABLE LAW, THE LICENSED WORK IS PROVIDED ON
67
+ AN “AS IS” BASIS. LICENSOR HEREBY DISCLAIMS ALL WARRANTIES AND CONDITIONS,
68
+ EXPRESS OR IMPLIED, INCLUDING (WITHOUT LIMITATION) WARRANTIES OF
69
+ MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, NON-INFRINGEMENT, AND
70
+ TITLE.
71
+
72
+ MariaDB hereby grants you permission to use this License’s text to license
73
+ your works, and to refer to it using the trademark “Business Source License”,
74
+ as long as you comply with the Covenants of Licensor below.
75
+
76
+ Covenants of Licensor
77
+
78
+ In consideration of the right to use this License’s text and the “Business
79
+ Source License” name and trademark, Licensor covenants to MariaDB, and to all
80
+ other recipients of the licensed work to be provided by Licensor:
81
+
82
+ 1. To specify as the Change License the GPL Version 2.0 or any later version,
83
+ or a license that is compatible with GPL Version 2.0 or a later version,
84
+ where “compatible” means that software provided under the Change License can
85
+ be included in a program with software provided under GPL Version 2.0 or a
86
+ later version. Licensor may specify additional Change Licenses without
87
+ limitation.
88
+
89
+ 2. To either: (a) specify an additional grant of rights to use that does not
90
+ impose any additional restriction on the right granted in this License, as
91
+ the Additional Use Grant; or (b) insert the text “None”.
92
+
93
+ 3. To specify a Change Date.
94
+
95
+ 4. Not to modify this License in any other way.
zerodom-0.0.1/PKG-INFO ADDED
@@ -0,0 +1,168 @@
1
+ Metadata-Version: 2.4
2
+ Name: zerodom
3
+ Version: 0.0.1
4
+ Summary: DOM-to-Interaction-Graph middleware for AI web agents
5
+ Project-URL: Homepage, https://github.com/DevHusnainAi/zerodom
6
+ Project-URL: Repository, https://github.com/DevHusnainAi/zerodom
7
+ Project-URL: Issues, https://github.com/DevHusnainAi/zerodom/issues
8
+ Author-email: Syed Husnain Khalid <contact@vexralabs.com>
9
+ License-Expression: BUSL-1.1
10
+ License-File: LICENSE
11
+ Keywords: agent,browser-automation,dom,llm,mcp,playwright,tokens
12
+ Classifier: Development Status :: 4 - Beta
13
+ Classifier: Intended Audience :: Developers
14
+ Classifier: License :: Other/Proprietary License
15
+ Classifier: Programming Language :: Python :: 3 :: Only
16
+ Classifier: Topic :: Internet :: WWW/HTTP :: Browsers
17
+ Classifier: Topic :: Text Processing :: Markup :: HTML
18
+ Requires-Python: >=3.10
19
+ Requires-Dist: lxml
20
+ Requires-Dist: mcp[cli]
21
+ Requires-Dist: playwright
22
+ Requires-Dist: tiktoken
23
+ Description-Content-Type: text/markdown
24
+
25
+ # ZeroDOM
26
+
27
+ ### Your agent doesn't need the DOM. It needs to know what it can click.
28
+
29
+ ZeroDOM turns a bloated HTML page into a token-optimized **interaction graph** —
30
+ the buttons, inputs and links an agent can actually act on, and nothing else.
31
+
32
+ An agent driving a browser gets one of two action spaces today, and both are bad.
33
+ **Pixels** are slow, expensive, and produce coordinates that go stale the moment
34
+ the page scrolls. **The accessibility tree** is cheaper but enormous, and it has
35
+ no stable handles: 102 of Hacker News' 220 actionable nodes share a
36
+ `(role, name)` pair with another node, so there is no way to say *which* story to
37
+ upvote.
38
+
39
+ ZeroDOM is a third option — a flat list of what the page can do, where every
40
+ entry has a stable id, and the addressing information that makes it clickable
41
+ never enters the context window. About **10 tokens per action, on every page
42
+ tested**.
43
+
44
+ - **No LLM in the loop.** lxml in, graph out, 10–30ms, identical output every run.
45
+ - **Selectors never enter the context window.** The model sees `[03]`; the CSS
46
+ path stays in `selector_map()` on your side.
47
+ - **Nothing leaves your machine.** No telemetry, no API keys, no storage — the
48
+ only network traffic is the page you pointed it at.
49
+ - **The selectors actually resolve.** Verified in a real browser: 231/231 on
50
+ Hacker News, where rows carry no `id` or `class` and numeric ids need escaping.
51
+
52
+ ## Install
53
+
54
+ ```bash
55
+ pip install zerodom
56
+ # or: uvx zerodom — the CLI runs straight off PyPI
57
+ ```
58
+
59
+ One extra step only if you use the browser-backed features (`from_page`,
60
+ `--render`, `--screenshot`, `--html`):
61
+
62
+ ```bash
63
+ playwright install chromium
64
+ ```
65
+
66
+ ## Quickstart
67
+
68
+ ```python
69
+ from zerodom import ZeroDOM
70
+
71
+ graph = ZeroDOM.from_page(page) # any Playwright page, sync or async
72
+ print(graph.to_compact_text()) # what you send the model
73
+ selectors = graph.selector_map() # {"node_01": "#email-input", ...} — stays your side
74
+ ```
75
+
76
+ Real output:
77
+
78
+ ```text
79
+ PAGE: Orbit — Fleet Console | file:///…/demo/demo-page.html
80
+ [01] a 'Fleet'
81
+ [05] input 'Search'
82
+ [07] input* 'Assigned driver' ph='Search by name'
83
+ [15] button 'Dispatch'
84
+ [17] button! 'Recall (in transit)'
85
+ ```
86
+
87
+ `*` marks required, `!` marks disabled. The model answers `click 15`. You resolve
88
+ `node_15` against `selector_map()` and click it.
89
+
90
+ ## Cost per action
91
+
92
+ The fair comparison isn't raw HTML — nobody sends a model raw HTML. It's
93
+ Playwright's `page.aria_snapshot(mode="ai")`, which is what Playwright MCP puts
94
+ in a model's context.
95
+
96
+ | page | ZeroDOM | ARIA |
97
+ |---|---:|---:|
98
+ | airbnb.com | **9.9** | 23.3 |
99
+ | github.com/…/issues | **12.1** | 70.2 |
100
+ | en.wikipedia.org article | **11.6** | 41.0 |
101
+ | news.ycombinator.com | **10.2** | 47.0 |
102
+ | developer.mozilla.org | **9.9** | 46.8 |
103
+
104
+ *tokens per actionable node — measured 2026-08-03*
105
+
106
+ **~10 tokens per action, flat across every page**, against 23–70 and wildly
107
+ variable. Mean 71.0% fewer tokens than the snapshot a model actually gets, and
108
+ 93.1% against raw HTML.
109
+
110
+ ## MCP server
111
+
112
+ ```json
113
+ {
114
+ "mcpServers": {
115
+ "zerodom": {
116
+ "command": "uvx",
117
+ "args": ["--from", "zerodom", "zerodom-mcp"]
118
+ }
119
+ }
120
+ }
121
+ ```
122
+
123
+ Goes in `claude_desktop_config.json` for Claude Desktop, or `.cursor/mcp.json`
124
+ for Cursor.
125
+
126
+ | tool | what it does |
127
+ |---|---|
128
+ | `zerodom_parse_url(url, verbose=False)` | navigate, return the compact graph |
129
+ | `zerodom_read_page(verbose=False)` | re-read the live DOM **without navigating** |
130
+ | `zerodom_find(query)` | return only the nodes matching a phrase |
131
+ | `zerodom_click_node(node_id)` | click, then return **what changed** |
132
+ | `zerodom_fill_node(node_id, text)` | type, then return what changed |
133
+
134
+ An agent loop shouldn't re-read the page it already has. `zerodom_find` answers
135
+ "where's the dispatch button?" in one line, and actions return a diff — `+`
136
+ appeared, `-` gone, `~` value changed — instead of re-listing every node.
137
+
138
+ ## CLI
139
+
140
+ ```bash
141
+ zerodom https://news.ycombinator.com # compact graph + token savings
142
+ zerodom https://example.com --find checkout # only the nodes that match
143
+ zerodom https://example.com --render # headless Chromium, for JS-heavy pages
144
+ zerodom https://example.com --json # full JSON graph, selectors included
145
+ zerodom https://example.com --html out.html # graph beside an annotated screenshot
146
+ ```
147
+
148
+ ## Limitations
149
+
150
+ - **Closed shadow roots** are unreachable — no browser API exposes them. Open
151
+ roots work.
152
+ - **Cross-origin and same-origin iframes** are not traversed yet; only the top
153
+ frame is parsed.
154
+ - **Canvas-rendered UIs** have no DOM to read. Use a vision model there.
155
+ - **Labels come from the page**, so a hostile page can write anything into one.
156
+ Treat graph text as untrusted input — see SECURITY.md.
157
+ - ZeroDOM operates on a `Page` you already control. It does not bypass bot
158
+ detection and makes no attempt to.
159
+
160
+ ## Links
161
+
162
+ - **Source, full README and benchmarks:**
163
+ https://github.com/DevHusnainAi/zerodom
164
+ - **Issues:** https://github.com/DevHusnainAi/zerodom/issues
165
+ - **License:** Business Source License 1.1 — free to use, including in
166
+ production; you may not offer ZeroDOM itself as a hosted service. Converts to
167
+ Apache 2.0 on 2030-08-09.
168
+